Skip to content

P
Paddle

Crayon鑫 / Paddle
与 Fork 源项目一致

Fork自 PaddlePaddle / Paddle

通知 1
Star 1
Fork 0
P
Paddle
未验证 提交 1cd7d149 编写于 作者: X xiaoting 提交者： GitHub
浏览文件
操作

fix typo for interp_v2,test=develop (#26843) (#27158) 

* fix typo for interp_v2,test=develop

* align with torch, test=develop

* add area mode, test=develop

* fix bug, test=develop

* format notes, test=develop

* update for converage, test=develop

* fix bilinear, test=develop

* fix bicubic, test=develop

* fix typo, test=develop

* fix coverage, test=develop

* fix helper.input_dtype, test=develop

* polish notes, test=develop

* polish notes, test=develop

* polish notes, test=develop
上级 73dbab3f
显示空白变更内容
内联 并排
Showing with 350 addition and 125 deletion
paddle/fluid/operators/interpolate_v2_op.cc
浏览文件 @ 1cd7d149
...@@ -67,7 +67,7 @@ static void Interpolate1DInferShapeCheck(framework::InferShapeContext* ctx) { ...@@ -67,7 +67,7 @@ static void Interpolate1DInferShapeCheck(framework::InferShapeContext* ctx) {
scale_tensor[0], 1, scale_tensor[0], 1,
platform::errors::InvalidArgument( platform::errors::InvalidArgument(
"Scale's shape must be 1, but got shape = %d .", scale_tensor[0])); "Scale's shape must be 1, but got shape = %d .", scale_tensor[0]));
// out_w = -1; out_w = -1;
} else { } else {
auto scale = ctx->Attrs().Get<std::vector<float>>("scale"); auto scale = ctx->Attrs().Get<std::vector<float>>("scale");
if (scale.size() > 0) { if (scale.size() > 0) {
...@@ -159,8 +159,8 @@ static void Interpolate2DInferShapeCheck(framework::InferShapeContext* ctx) { ...@@ -159,8 +159,8 @@ static void Interpolate2DInferShapeCheck(framework::InferShapeContext* ctx) {
platform::errors::InvalidArgument( platform::errors::InvalidArgument(
"Scale's shape must be 2 or 1, but got shape = %d .", "Scale's shape must be 2 or 1, but got shape = %d .",
scale_tensor[0])); scale_tensor[0]));
// out_h = -1; out_h = -1;
// out_w = -1; out_w = -1;
} else { } else {
auto scale = ctx->Attrs().Get<std::vector<float>>("scale"); auto scale = ctx->Attrs().Get<std::vector<float>>("scale");
if (scale.size() > 0) { if (scale.size() > 0) {
...@@ -264,9 +264,9 @@ static void Interpolate3DInferShapeCheck(framework::InferShapeContext* ctx) { ...@@ -264,9 +264,9 @@ static void Interpolate3DInferShapeCheck(framework::InferShapeContext* ctx) {
platform::errors::InvalidArgument( platform::errors::InvalidArgument(
"Scale's shape must be 3 or 1, but got shape = %d .", "Scale's shape must be 3 or 1, but got shape = %d .",
scale_tensor[0])); scale_tensor[0]));
// out_d = -1; out_d = -1;
// out_h = -1; out_h = -1;
// out_w = -1; out_w = -1;
} else { } else {
auto scale = ctx->Attrs().Get<std::vector<float>>("scale"); auto scale = ctx->Attrs().Get<std::vector<float>>("scale");
if (scale.size() > 0) { if (scale.size() > 0) {
...@@ -633,6 +633,9 @@ DECLARE_NO_NEED_BUFFER_VARS_INFERER(InterpolateV2GradNoNeedBufferVarsInferer, ...@@ -633,6 +633,9 @@ DECLARE_NO_NEED_BUFFER_VARS_INFERER(InterpolateV2GradNoNeedBufferVarsInferer,
} // namespace operators } // namespace operators
} // namespace paddle } // namespace paddle
// interp_v2 support scale_factor whose input type is list, this operation is
// not
// compatible with interp_op, so a new one is added in paddle2.0
namespace ops = paddle::operators; namespace ops = paddle::operators;
REGISTER_OPERATOR(bilinear_interp_v2, ops::InterpolateV2Op, REGISTER_OPERATOR(bilinear_interp_v2, ops::InterpolateV2Op,
ops::InterpolateV2OpMaker, ops::InterpolateV2OpMaker,
......
paddle/fluid/operators/interpolate_v2_op.cu
浏览文件 @ 1cd7d149
...@@ -836,12 +836,12 @@ static void Interpolate1DCUDAFwd(const framework::ExecutionContext& ctx, ...@@ -836,12 +836,12 @@ static void Interpolate1DCUDAFwd(const framework::ExecutionContext& ctx,
int out_w = ctx.Attr<int>("out_w"); int out_w = ctx.Attr<int>("out_w");
auto list_new_shape_tensor = ctx.MultiInput<framework::Tensor>("SizeTensor"); auto list_new_shape_tensor = ctx.MultiInput<framework::Tensor>("SizeTensor");
float scale_w = -1;
if (list_new_shape_tensor.size() > 0) { if (list_new_shape_tensor.size() > 0) {
// have size tensor // have size tensor
auto new_size = get_new_shape(list_new_shape_tensor); auto new_size = get_new_shape(list_new_shape_tensor);
out_w = new_size[0]; out_w = new_size[0];
} else { } else {
float scale_w = -1;
auto scale_tensor = ctx.Input<Tensor>("Scale"); auto scale_tensor = ctx.Input<Tensor>("Scale");
auto scale = ctx.Attr<std::vector<float>>("scale"); auto scale = ctx.Attr<std::vector<float>>("scale");
if (scale_tensor != nullptr) { if (scale_tensor != nullptr) {
...@@ -887,8 +887,11 @@ static void Interpolate1DCUDAFwd(const framework::ExecutionContext& ctx, ...@@ -887,8 +887,11 @@ static void Interpolate1DCUDAFwd(const framework::ExecutionContext& ctx,
float ratio_w = 0.f; float ratio_w = 0.f;
if (out_w > 1) { if (out_w > 1) {
ratio_w = (align_corners) ? static_cast<float>(in_w - 1.0) / (out_w - 1.0) float new_scale_w = 0.f;
new_scale_w = (scale_w > 0) ? static_cast<float>(1. / scale_w)
: static_cast<float>(in_w) / out_w; : static_cast<float>(in_w) / out_w;
ratio_w = (align_corners) ? static_cast<float>(in_w - 1.0) / (out_w - 1.0)
: static_cast<float>(new_scale_w);
} }
int in_cw = c * in_w; int in_cw = c * in_w;
...@@ -924,14 +927,14 @@ static void Interpolate2DCUDAFwd(const framework::ExecutionContext& ctx, ...@@ -924,14 +927,14 @@ static void Interpolate2DCUDAFwd(const framework::ExecutionContext& ctx,
int out_w = ctx.Attr<int>("out_w"); int out_w = ctx.Attr<int>("out_w");
auto list_new_shape_tensor = ctx.MultiInput<framework::Tensor>("SizeTensor"); auto list_new_shape_tensor = ctx.MultiInput<framework::Tensor>("SizeTensor");
float scale_w = -1;
float scale_h = -1;
if (list_new_shape_tensor.size() > 0) { if (list_new_shape_tensor.size() > 0) {
// have size tensor // have size tensor
auto new_size = get_new_shape(list_new_shape_tensor); auto new_size = get_new_shape(list_new_shape_tensor);
out_h = new_size[0]; out_h = new_size[0];
out_w = new_size[1]; out_w = new_size[1];
} else { } else {
float scale_h = -1;
float scale_w = -1;
auto scale_tensor = ctx.Input<Tensor>("Scale"); auto scale_tensor = ctx.Input<Tensor>("Scale");
auto scale = ctx.Attr<std::vector<float>>("scale"); auto scale = ctx.Attr<std::vector<float>>("scale");
if (scale_tensor != nullptr) { if (scale_tensor != nullptr) {
...@@ -993,12 +996,18 @@ static void Interpolate2DCUDAFwd(const framework::ExecutionContext& ctx, ...@@ -993,12 +996,18 @@ static void Interpolate2DCUDAFwd(const framework::ExecutionContext& ctx,
float ratio_h = 0.f; float ratio_h = 0.f;
float ratio_w = 0.f; float ratio_w = 0.f;
if (out_h > 1) { if (out_h > 1) {
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1) float new_scale_h = 0.f;
new_scale_h = (scale_h > 0) ? static_cast<float>(1. / scale_h)
: static_cast<float>(in_h) / out_h; : static_cast<float>(in_h) / out_h;
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1)
: static_cast<float>(new_scale_h);
} }
if (out_w > 1) { if (out_w > 1) {
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1) float new_scale_w = 0.f;
new_scale_w = (scale_w > 0) ? static_cast<float>(1. / scale_w)
: static_cast<float>(in_w) / out_w; : static_cast<float>(in_w) / out_w;
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1)
: static_cast<float>(new_scale_w);
} }
int in_hw = in_h * in_w; int in_hw = in_h * in_w;
...@@ -1048,6 +1057,9 @@ static void Interpolate3DCUDAFwd(const framework::ExecutionContext& ctx, ...@@ -1048,6 +1057,9 @@ static void Interpolate3DCUDAFwd(const framework::ExecutionContext& ctx,
int out_w = ctx.Attr<int>("out_w"); int out_w = ctx.Attr<int>("out_w");
auto list_new_shape_tensor = ctx.MultiInput<framework::Tensor>("SizeTensor"); auto list_new_shape_tensor = ctx.MultiInput<framework::Tensor>("SizeTensor");
float scale_w = -1;
float scale_d = -1;
float scale_h = -1;
if (list_new_shape_tensor.size() > 0) { if (list_new_shape_tensor.size() > 0) {
// have size tensor // have size tensor
auto new_size = get_new_shape(list_new_shape_tensor); auto new_size = get_new_shape(list_new_shape_tensor);
...@@ -1055,9 +1067,6 @@ static void Interpolate3DCUDAFwd(const framework::ExecutionContext& ctx, ...@@ -1055,9 +1067,6 @@ static void Interpolate3DCUDAFwd(const framework::ExecutionContext& ctx,
out_h = new_size[1]; out_h = new_size[1];
out_w = new_size[2]; out_w = new_size[2];
} else { } else {
float scale_d = -1;
float scale_h = -1;
float scale_w = -1;
auto scale_tensor = ctx.Input<Tensor>("Scale"); auto scale_tensor = ctx.Input<Tensor>("Scale");
auto scale = ctx.Attr<std::vector<float>>("scale"); auto scale = ctx.Attr<std::vector<float>>("scale");
if (scale_tensor != nullptr) { if (scale_tensor != nullptr) {
...@@ -1129,16 +1138,25 @@ static void Interpolate3DCUDAFwd(const framework::ExecutionContext& ctx, ...@@ -1129,16 +1138,25 @@ static void Interpolate3DCUDAFwd(const framework::ExecutionContext& ctx,
float ratio_h = 0.f; float ratio_h = 0.f;
float ratio_w = 0.f; float ratio_w = 0.f;
if (out_d > 1) { if (out_d > 1) {
ratio_d = (align_corners) ? static_cast<float>(in_d - 1) / (out_d - 1) float new_scale_d = 0.f;
new_scale_d = (scale_d > 0) ? static_cast<float>(1. / scale_d)
: static_cast<float>(in_d) / out_d; : static_cast<float>(in_d) / out_d;
ratio_d = (align_corners) ? static_cast<float>(in_d - 1) / (out_d - 1)
: static_cast<float>(new_scale_d);
} }
if (out_h > 1) { if (out_h > 1) {
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1) float new_scale_h = 0.f;
new_scale_h = (scale_h > 0) ? static_cast<float>(1. / scale_h)
: static_cast<float>(in_h) / out_h; : static_cast<float>(in_h) / out_h;
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1)
: static_cast<float>(new_scale_h);
} }
if (out_w > 1) { if (out_w > 1) {
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1) float new_scale_w = 0.f;
new_scale_w = (scale_w > 0) ? static_cast<float>(1. / scale_w)
: static_cast<float>(in_w) / out_w; : static_cast<float>(in_w) / out_w;
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1)
: static_cast<float>(new_scale_w);
} }
int in_dhw = in_d * in_h * in_w; int in_dhw = in_d * in_h * in_w;
...@@ -1230,8 +1248,11 @@ static void Interpolate1DCUDABwd(const framework::ExecutionContext& ctx, ...@@ -1230,8 +1248,11 @@ static void Interpolate1DCUDABwd(const framework::ExecutionContext& ctx,
float ratio_w = 0.f; float ratio_w = 0.f;
if (out_w > 1) { if (out_w > 1) {
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1) float new_scale_w = 0.f;
new_scale_w = (scale_w > 0) ? static_cast<float>(1. / scale_w)
: static_cast<float>(in_w) / out_w; : static_cast<float>(in_w) / out_w;
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1)
: static_cast<float>(new_scale_w);
} }
int in_cw = c * in_w; int in_cw = c * in_w;
int out_cw = c * out_w; int out_cw = c * out_w;
...@@ -1333,12 +1354,18 @@ static void Interpolate2DCUDABwd(const framework::ExecutionContext& ctx, ...@@ -1333,12 +1354,18 @@ static void Interpolate2DCUDABwd(const framework::ExecutionContext& ctx,
float ratio_h = 0.f; float ratio_h = 0.f;
float ratio_w = 0.f; float ratio_w = 0.f;
if (out_h > 1) { if (out_h > 1) {
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1) float new_scale_h = 0.f;
new_scale_h = (scale_h > 0) ? static_cast<float>(1. / scale_h)
: static_cast<float>(in_h) / out_h; : static_cast<float>(in_h) / out_h;
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1)
: static_cast<float>(new_scale_h);
} }
if (out_w > 1) { if (out_w > 1) {
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1) float new_scale_w = 0.f;
new_scale_w = (scale_w > 0) ? static_cast<float>(1. / scale_w)
: static_cast<float>(in_w) / out_w; : static_cast<float>(in_w) / out_w;
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1)
: static_cast<float>(new_scale_w);
} }
int in_hw = in_h * in_w; int in_hw = in_h * in_w;
...@@ -1464,16 +1491,25 @@ static void Interpolate3DCUDABwd(const framework::ExecutionContext& ctx, ...@@ -1464,16 +1491,25 @@ static void Interpolate3DCUDABwd(const framework::ExecutionContext& ctx,
float ratio_h = 0.f; float ratio_h = 0.f;
float ratio_w = 0.f; float ratio_w = 0.f;
if (out_d > 1) { if (out_d > 1) {
ratio_d = (align_corners) ? static_cast<float>(in_d - 1) / (out_d - 1) float new_scale_d = 0.f;
new_scale_d = (scale_d > 0) ? static_cast<float>(1. / scale_d)
: static_cast<float>(in_d) / out_d; : static_cast<float>(in_d) / out_d;
ratio_d = (align_corners) ? static_cast<float>(in_d - 1) / (out_d - 1)
: static_cast<float>(new_scale_d);
} }
if (out_h > 1) { if (out_h > 1) {
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1) float new_scale_h = 0.f;
new_scale_h = (scale_h > 0) ? static_cast<float>(1. / scale_h)
: static_cast<float>(in_h) / out_h; : static_cast<float>(in_h) / out_h;
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1)
: static_cast<float>(new_scale_h);
} }
if (out_w > 1) { if (out_w > 1) {
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1) float new_scale_w = 0.f;
new_scale_w = (scale_w > 0) ? static_cast<float>(1. / scale_w)
: static_cast<float>(in_w) / out_w; : static_cast<float>(in_w) / out_w;
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1)
: static_cast<float>(new_scale_w);
} }
int in_dhw = in_d * in_h * in_w; int in_dhw = in_d * in_h * in_w;
......
paddle/fluid/operators/interpolate_v2_op.h
浏览文件 @ 1cd7d149
...@@ -783,12 +783,13 @@ static void Interpolate1DCPUFwd(const framework::ExecutionContext& ctx, ...@@ -783,12 +783,13 @@ static void Interpolate1DCPUFwd(const framework::ExecutionContext& ctx,
int out_w = ctx.Attr<int>("out_w"); int out_w = ctx.Attr<int>("out_w");
auto list_new_size_tensor = ctx.MultiInput<framework::Tensor>("SizeTensor"); auto list_new_size_tensor = ctx.MultiInput<framework::Tensor>("SizeTensor");
float scale_w = -1.;
if (list_new_size_tensor.size() > 0) { if (list_new_size_tensor.size() > 0) {
// have size tensor // have size tensor
auto new_size = get_new_shape(list_new_size_tensor); auto new_size = get_new_shape(list_new_size_tensor);
out_w = new_size[0]; out_w = new_size[0];
} else { } else {
float scale_w = -1; // float scale_w = -1;
auto scale_tensor = ctx.Input<Tensor>("Scale"); auto scale_tensor = ctx.Input<Tensor>("Scale");
auto scale = ctx.Attr<std::vector<float>>("scale"); auto scale = ctx.Attr<std::vector<float>>("scale");
if (scale_tensor != nullptr) { if (scale_tensor != nullptr) {
...@@ -833,8 +834,11 @@ static void Interpolate1DCPUFwd(const framework::ExecutionContext& ctx, ...@@ -833,8 +834,11 @@ static void Interpolate1DCPUFwd(const framework::ExecutionContext& ctx,
float ratio_w = 0.f; float ratio_w = 0.f;
if (out_w > 1) { if (out_w > 1) {
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1) float new_scale_w = 0.f;
new_scale_w = (scale_w > 0) ? static_cast<float>(1. / scale_w)
: static_cast<float>(in_w) / out_w; : static_cast<float>(in_w) / out_w;
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1)
: static_cast<float>(new_scale_w);
} }
if ("linear" == interp_method) { if ("linear" == interp_method) {
LinearInterpolation<T>(input, output, ratio_w, in_w, n, c, out_w, LinearInterpolation<T>(input, output, ratio_w, in_w, n, c, out_w,
...@@ -856,6 +860,8 @@ static void Interpolate2DCPUFwd(const framework::ExecutionContext& ctx, ...@@ -856,6 +860,8 @@ static void Interpolate2DCPUFwd(const framework::ExecutionContext& ctx,
int out_h = ctx.Attr<int>("out_h"); int out_h = ctx.Attr<int>("out_h");
int out_w = ctx.Attr<int>("out_w"); int out_w = ctx.Attr<int>("out_w");
float scale_h = -1;
float scale_w = -1;
auto list_new_size_tensor = ctx.MultiInput<framework::Tensor>("SizeTensor"); auto list_new_size_tensor = ctx.MultiInput<framework::Tensor>("SizeTensor");
if (list_new_size_tensor.size() > 0) { if (list_new_size_tensor.size() > 0) {
...@@ -864,8 +870,6 @@ static void Interpolate2DCPUFwd(const framework::ExecutionContext& ctx, ...@@ -864,8 +870,6 @@ static void Interpolate2DCPUFwd(const framework::ExecutionContext& ctx,
out_h = new_size[0]; out_h = new_size[0];
out_w = new_size[1]; out_w = new_size[1];
} else { } else {
float scale_h = -1;
float scale_w = -1;
auto scale_tensor = ctx.Input<Tensor>("Scale"); auto scale_tensor = ctx.Input<Tensor>("Scale");
auto scale = ctx.Attr<std::vector<float>>("scale"); auto scale = ctx.Attr<std::vector<float>>("scale");
if (scale_tensor != nullptr) { if (scale_tensor != nullptr) {
...@@ -925,12 +929,18 @@ static void Interpolate2DCPUFwd(const framework::ExecutionContext& ctx, ...@@ -925,12 +929,18 @@ static void Interpolate2DCPUFwd(const framework::ExecutionContext& ctx,
float ratio_h = 0.f; float ratio_h = 0.f;
float ratio_w = 0.f; float ratio_w = 0.f;
if (out_h > 1) { if (out_h > 1) {
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1) float new_scale_h = 0.f;
new_scale_h = (scale_h > 0) ? static_cast<float>(1. / scale_h)
: static_cast<float>(in_h) / out_h; : static_cast<float>(in_h) / out_h;
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1)
: static_cast<float>(new_scale_h);
} }
if (out_w > 1) { if (out_w > 1) {
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1) float new_scale_w = 0.f;
new_scale_w = (scale_w > 0) ? static_cast<float>(1. / scale_w)
: static_cast<float>(in_w) / out_w; : static_cast<float>(in_w) / out_w;
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1)
: static_cast<float>(new_scale_w);
} }
if ("bilinear" == interp_method) { if ("bilinear" == interp_method) {
...@@ -962,6 +972,10 @@ static void Interpolate3DCPUFwd(const framework::ExecutionContext& ctx, ...@@ -962,6 +972,10 @@ static void Interpolate3DCPUFwd(const framework::ExecutionContext& ctx,
int out_h = ctx.Attr<int>("out_h"); int out_h = ctx.Attr<int>("out_h");
int out_w = ctx.Attr<int>("out_w"); int out_w = ctx.Attr<int>("out_w");
float scale_d = -1;
float scale_h = -1;
float scale_w = -1;
auto list_new_size_tensor = ctx.MultiInput<framework::Tensor>("SizeTensor"); auto list_new_size_tensor = ctx.MultiInput<framework::Tensor>("SizeTensor");
if (list_new_size_tensor.size() > 0) { if (list_new_size_tensor.size() > 0) {
// have size tensor // have size tensor
...@@ -970,9 +984,6 @@ static void Interpolate3DCPUFwd(const framework::ExecutionContext& ctx, ...@@ -970,9 +984,6 @@ static void Interpolate3DCPUFwd(const framework::ExecutionContext& ctx,
out_h = new_size[1]; out_h = new_size[1];
out_w = new_size[2]; out_w = new_size[2];
} else { } else {
float scale_d = -1;
float scale_h = -1;
float scale_w = -1;
auto scale_tensor = ctx.Input<Tensor>("Scale"); auto scale_tensor = ctx.Input<Tensor>("Scale");
auto scale = ctx.Attr<std::vector<float>>("scale"); auto scale = ctx.Attr<std::vector<float>>("scale");
if (scale_tensor != nullptr) { if (scale_tensor != nullptr) {
...@@ -1043,16 +1054,25 @@ static void Interpolate3DCPUFwd(const framework::ExecutionContext& ctx, ...@@ -1043,16 +1054,25 @@ static void Interpolate3DCPUFwd(const framework::ExecutionContext& ctx,
float ratio_h = 0.f; float ratio_h = 0.f;
float ratio_w = 0.f; float ratio_w = 0.f;
if (out_d > 1) { if (out_d > 1) {
ratio_d = (align_corners) ? static_cast<float>(in_d - 1) / (out_d - 1) float new_scale_d = 0.f;
new_scale_d = (scale_d > 0) ? static_cast<float>(1. / scale_d)
: static_cast<float>(in_d) / out_d; : static_cast<float>(in_d) / out_d;
ratio_d = (align_corners) ? static_cast<float>(in_d - 1) / (out_d - 1)
: static_cast<float>(new_scale_d);
} }
if (out_h > 1) { if (out_h > 1) {
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1) float new_scale_h = 0.f;
new_scale_h = (scale_h > 0) ? static_cast<float>(1. / scale_h)
: static_cast<float>(in_h) / out_h; : static_cast<float>(in_h) / out_h;
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1)
: static_cast<float>(new_scale_h);
} }
if (out_w > 1) { if (out_w > 1) {
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1) float new_scale_w = 0.f;
new_scale_w = (scale_w > 0) ? static_cast<float>(1. / scale_w)
: static_cast<float>(in_w) / out_w; : static_cast<float>(in_w) / out_w;
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1)
: static_cast<float>(new_scale_w);
} }
if ("trilinear" == interp_method) { if ("trilinear" == interp_method) {
...@@ -1127,8 +1147,11 @@ static void Interpolate1DCPUBwd(const framework::ExecutionContext& ctx, ...@@ -1127,8 +1147,11 @@ static void Interpolate1DCPUBwd(const framework::ExecutionContext& ctx,
float ratio_w = 0.f; float ratio_w = 0.f;
if (out_w > 1) { if (out_w > 1) {
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1) float new_scale_w = 0.f;
new_scale_w = (scale_w > 0) ? static_cast<float>(1. / scale_w)
: static_cast<float>(in_w) / out_w; : static_cast<float>(in_w) / out_w;
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1)
: static_cast<float>(new_scale_w);
} }
if ("linear" == interp_method) { if ("linear" == interp_method) {
LinearInterpolationGrad<T>(output_grad, input_grad, ratio_w, in_w, n, c, LinearInterpolationGrad<T>(output_grad, input_grad, ratio_w, in_w, n, c,
...@@ -1216,12 +1239,18 @@ static void Interpolate2DCPUBwd(const framework::ExecutionContext& ctx, ...@@ -1216,12 +1239,18 @@ static void Interpolate2DCPUBwd(const framework::ExecutionContext& ctx,
float ratio_h = 0.f; float ratio_h = 0.f;
float ratio_w = 0.f; float ratio_w = 0.f;
if (out_h > 1) { if (out_h > 1) {
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1) float new_scale_h = 0.f;
new_scale_h = (scale_h > 0) ? static_cast<float>(1. / scale_h)
: static_cast<float>(in_h) / out_h; : static_cast<float>(in_h) / out_h;
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1)
: static_cast<float>(new_scale_h);
} }
if (out_w > 1) { if (out_w > 1) {
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1) float new_scale_w = 0.f;
new_scale_w = (scale_w > 0) ? static_cast<float>(1. / scale_w)
: static_cast<float>(in_w) / out_w; : static_cast<float>(in_w) / out_w;
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1)
: static_cast<float>(new_scale_w);
} }
if ("bilinear" == interp_method) { if ("bilinear" == interp_method) {
...@@ -1327,16 +1356,25 @@ static void Interpolate3DCPUBwd(const framework::ExecutionContext& ctx, ...@@ -1327,16 +1356,25 @@ static void Interpolate3DCPUBwd(const framework::ExecutionContext& ctx,
float ratio_h = 0.f; float ratio_h = 0.f;
float ratio_w = 0.f; float ratio_w = 0.f;
if (out_d > 1) { if (out_d > 1) {
ratio_d = (align_corners) ? static_cast<float>(in_d - 1) / (out_d - 1) float new_scale_d = 0.f;
new_scale_d = (scale_d > 0) ? static_cast<float>(1. / scale_d)
: static_cast<float>(in_d) / out_d; : static_cast<float>(in_d) / out_d;
ratio_d = (align_corners) ? static_cast<float>(in_d - 1) / (out_d - 1)
: static_cast<float>(new_scale_d);
} }
if (out_h > 1) { if (out_h > 1) {
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1) float new_scale_h = 0.f;
new_scale_h = (scale_h > 0) ? static_cast<float>(1. / scale_h)
: static_cast<float>(in_h) / out_h; : static_cast<float>(in_h) / out_h;
ratio_h = (align_corners) ? static_cast<float>(in_h - 1) / (out_h - 1)
: static_cast<float>(new_scale_h);
} }
if (out_w > 1) { if (out_w > 1) {
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1) float new_scale_w = 0.f;
new_scale_w = (scale_w > 0) ? static_cast<float>(1. / scale_w)
: static_cast<float>(in_w) / out_w; : static_cast<float>(in_w) / out_w;
ratio_w = (align_corners) ? static_cast<float>(in_w - 1) / (out_w - 1)
: static_cast<float>(new_scale_w);
} }
if ("trilinear" == interp_method) { if ("trilinear" == interp_method) {
......
python/paddle/fluid/tests/unittests/test_adaptive_avg_pool1d.py
浏览文件 @ 1cd7d149
...@@ -94,6 +94,10 @@ class TestPool1d_API(unittest.TestCase): ...@@ -94,6 +94,10 @@ class TestPool1d_API(unittest.TestCase):
result = ada_max_pool1d_dg(input) result = ada_max_pool1d_dg(input)
self.assertTrue(np.allclose(result.numpy(), result_np)) self.assertTrue(np.allclose(result.numpy(), result_np))
result = paddle.nn.functional.common.interpolate(
input, mode="area", size=16)
self.assertTrue(np.allclose(result.numpy(), result_np))
def check_adaptive_avg_static_results(self, place): def check_adaptive_avg_static_results(self, place):
with fluid.program_guard(fluid.Program(), fluid.Program()): with fluid.program_guard(fluid.Program(), fluid.Program()):
input = fluid.data(name="input", shape=[2, 3, 32], dtype="float32") input = fluid.data(name="input", shape=[2, 3, 32], dtype="float32")
......
python/paddle/fluid/tests/unittests/test_adaptive_avg_pool2d.py
浏览文件 @ 1cd7d149
...@@ -163,6 +163,9 @@ class TestAdaptiveAvgPool2dAPI(unittest.TestCase): ...@@ -163,6 +163,9 @@ class TestAdaptiveAvgPool2dAPI(unittest.TestCase):
out_5 = paddle.nn.functional.adaptive_avg_pool2d( out_5 = paddle.nn.functional.adaptive_avg_pool2d(
x=x, output_size=[None, 3]) x=x, output_size=[None, 3])
out_6 = paddle.nn.functional.interpolate(
x=x, mode="area", size=[2, 5])
assert np.allclose(out_1.numpy(), self.res_1_np) assert np.allclose(out_1.numpy(), self.res_1_np)
assert np.allclose(out_2.numpy(), self.res_2_np) assert np.allclose(out_2.numpy(), self.res_2_np)
...@@ -173,6 +176,8 @@ class TestAdaptiveAvgPool2dAPI(unittest.TestCase): ...@@ -173,6 +176,8 @@ class TestAdaptiveAvgPool2dAPI(unittest.TestCase):
assert np.allclose(out_5.numpy(), self.res_5_np) assert np.allclose(out_5.numpy(), self.res_5_np)
assert np.allclose(out_6.numpy(), self.res_3_np)
class TestAdaptiveAvgPool2dClassAPI(unittest.TestCase): class TestAdaptiveAvgPool2dClassAPI(unittest.TestCase):
def setUp(self): def setUp(self):
......
python/paddle/fluid/tests/unittests/test_adaptive_avg_pool3d.py
浏览文件 @ 1cd7d149
...@@ -178,6 +178,9 @@ class TestAdaptiveAvgPool3dAPI(unittest.TestCase): ...@@ -178,6 +178,9 @@ class TestAdaptiveAvgPool3dAPI(unittest.TestCase):
out_5 = paddle.nn.functional.adaptive_avg_pool3d( out_5 = paddle.nn.functional.adaptive_avg_pool3d(
x=x, output_size=[None, 3, None]) x=x, output_size=[None, 3, None])
out_6 = paddle.nn.functional.interpolate(
x=x, mode="area", size=[2, 3, 5])
assert np.allclose(out_1.numpy(), self.res_1_np) assert np.allclose(out_1.numpy(), self.res_1_np)
assert np.allclose(out_2.numpy(), self.res_2_np) assert np.allclose(out_2.numpy(), self.res_2_np)
...@@ -188,6 +191,8 @@ class TestAdaptiveAvgPool3dAPI(unittest.TestCase): ...@@ -188,6 +191,8 @@ class TestAdaptiveAvgPool3dAPI(unittest.TestCase):
assert np.allclose(out_5.numpy(), self.res_5_np) assert np.allclose(out_5.numpy(), self.res_5_np)
assert np.allclose(out_6.numpy(), self.res_3_np)
class TestAdaptiveAvgPool3dClassAPI(unittest.TestCase): class TestAdaptiveAvgPool3dClassAPI(unittest.TestCase):
def setUp(self): def setUp(self):
......
python/paddle/fluid/tests/unittests/test_bicubic_interp_v2_op.py
浏览文件 @ 1cd7d149
...@@ -53,6 +53,8 @@ def value_bound(input, w, h, x, y): ...@@ -53,6 +53,8 @@ def value_bound(input, w, h, x, y):
def bicubic_interp_np(input, def bicubic_interp_np(input,
out_h, out_h,
out_w, out_w,
scale_h=0,
scale_w=0,
out_size=None, out_size=None,
actual_shape=None, actual_shape=None,
align_corners=True, align_corners=True,
...@@ -72,12 +74,18 @@ def bicubic_interp_np(input, ...@@ -72,12 +74,18 @@ def bicubic_interp_np(input,
if out_h > 1: if out_h > 1:
if (align_corners): if (align_corners):
ratio_h = (in_h - 1.0) / (out_h - 1.0) ratio_h = (in_h - 1.0) / (out_h - 1.0)
else:
if scale_h > 0:
ratio_h = 1.0 / scale_h
else: else:
ratio_h = 1.0 * in_h / out_h ratio_h = 1.0 * in_h / out_h
if out_w > 1: if out_w > 1:
if (align_corners): if (align_corners):
ratio_w = (in_w - 1.0) / (out_w - 1.0) ratio_w = (in_w - 1.0) / (out_w - 1.0)
else:
if scale_w > 0:
ratio_w = 1.0 / scale_w
else: else:
ratio_w = 1.0 * in_w / out_w ratio_w = 1.0 * in_w / out_w
...@@ -128,7 +136,8 @@ class TestBicubicInterpOp(OpTest): ...@@ -128,7 +136,8 @@ class TestBicubicInterpOp(OpTest):
self.init_test_case() self.init_test_case()
self.op_type = "bicubic_interp_v2" self.op_type = "bicubic_interp_v2"
input_np = np.random.random(self.input_shape).astype("float64") input_np = np.random.random(self.input_shape).astype("float64")
scale_h = 0
scale_w = 0
if self.data_layout == "NCHW": if self.data_layout == "NCHW":
in_h = self.input_shape[2] in_h = self.input_shape[2]
in_w = self.input_shape[3] in_w = self.input_shape[3]
...@@ -151,9 +160,9 @@ class TestBicubicInterpOp(OpTest): ...@@ -151,9 +160,9 @@ class TestBicubicInterpOp(OpTest):
out_h = self.out_h out_h = self.out_h
out_w = self.out_w out_w = self.out_w
output_np = bicubic_interp_np(input_np, out_h, out_w, self.out_size, output_np = bicubic_interp_np(input_np, out_h, out_w, scale_h, scale_w,
self.actual_shape, self.align_corners, self.out_size, self.actual_shape,
self.data_layout) self.align_corners, self.data_layout)
self.inputs = {'X': input_np} self.inputs = {'X': input_np}
if self.out_size is not None: if self.out_size is not None:
self.inputs['OutSize'] = self.out_size self.inputs['OutSize'] = self.out_size
...@@ -480,10 +489,34 @@ class TestBicubicOpError(unittest.TestCase): ...@@ -480,10 +489,34 @@ class TestBicubicOpError(unittest.TestCase):
out = interpolate( out = interpolate(
x, x,
size=None, size=None,
mode='trilinear', mode='bicubic',
align_corners=False, align_corners=False,
scale_factor=[1, 2, 2]) scale_factor=[1, 2, 2])
def test_size_and_scale():
x = fluid.data(name="x", shape=[2, 3, 6, 6], dtype="float32")
out = interpolate(
x,
size=None,
mode='bicubic',
align_corners=False,
scale_factor=None)
def test_size_and_scale2():
x = fluid.data(
name="input", shape=[2, 3, 6, 9, 4], dtype="float32")
out = interpolate(
x,
size=[2, 2, 2],
mode='trilinear',
align_corners=False,
scale_factor=2.0)
def test_size_type():
x = fluid.data(name="x", shape=[2, 3, 6, 6], dtype="float32")
out = interpolate(
x, size={2, 2}, mode='bicubic', align_corners=False)
self.assertRaises(ValueError, test_mode_type) self.assertRaises(ValueError, test_mode_type)
self.assertRaises(ValueError, test_input_shape) self.assertRaises(ValueError, test_input_shape)
self.assertRaises(TypeError, test_align_corcers) self.assertRaises(TypeError, test_align_corcers)
...@@ -498,6 +531,9 @@ class TestBicubicOpError(unittest.TestCase): ...@@ -498,6 +531,9 @@ class TestBicubicOpError(unittest.TestCase):
self.assertRaises(ValueError, test_align_corners_and_nearest) self.assertRaises(ValueError, test_align_corners_and_nearest)
self.assertRaises(ValueError, test_scale_shape) self.assertRaises(ValueError, test_scale_shape)
self.assertRaises(ValueError, test_scale_value) self.assertRaises(ValueError, test_scale_value)
self.assertRaises(ValueError, test_size_and_scale)
self.assertRaises(ValueError, test_size_and_scale2)
self.assertRaises(TypeError, test_size_type)
if __name__ == "__main__": if __name__ == "__main__":
......
python/paddle/fluid/tests/unittests/test_bilinear_interp_v2_op.py
浏览文件 @ 1cd7d149
...@@ -26,6 +26,8 @@ import paddle ...@@ -26,6 +26,8 @@ import paddle
def bilinear_interp_np(input, def bilinear_interp_np(input,
out_h, out_h,
out_w, out_w,
scale_w=0,
scale_h=0,
out_size=None, out_size=None,
actual_shape=None, actual_shape=None,
align_corners=True, align_corners=True,
...@@ -46,11 +48,17 @@ def bilinear_interp_np(input, ...@@ -46,11 +48,17 @@ def bilinear_interp_np(input,
if out_h > 1: if out_h > 1:
if (align_corners): if (align_corners):
ratio_h = (in_h - 1.0) / (out_h - 1.0) ratio_h = (in_h - 1.0) / (out_h - 1.0)
else:
if scale_h > 0:
ratio_h = 1.0 / scale_h
else: else:
ratio_h = 1.0 * in_h / out_h ratio_h = 1.0 * in_h / out_h
if out_w > 1: if out_w > 1:
if (align_corners): if (align_corners):
ratio_w = (in_w - 1.0) / (out_w - 1.0) ratio_w = (in_w - 1.0) / (out_w - 1.0)
else:
if scale_w > 0:
ratio_w = 1.0 / scale_w
else: else:
ratio_w = 1.0 * in_w / out_w ratio_w = 1.0 * in_w / out_w
...@@ -110,7 +118,8 @@ class TestBilinearInterpOp(OpTest): ...@@ -110,7 +118,8 @@ class TestBilinearInterpOp(OpTest):
else: else:
in_h = self.input_shape[1] in_h = self.input_shape[1]
in_w = self.input_shape[2] in_w = self.input_shape[2]
scale_h = 0
scale_w = 0
if self.scale: if self.scale:
if isinstance(self.scale, float) or isinstance(self.scale, int): if isinstance(self.scale, float) or isinstance(self.scale, int):
if self.scale > 0.: if self.scale > 0.:
...@@ -126,9 +135,9 @@ class TestBilinearInterpOp(OpTest): ...@@ -126,9 +135,9 @@ class TestBilinearInterpOp(OpTest):
out_h = self.out_h out_h = self.out_h
out_w = self.out_w out_w = self.out_w
output_np = bilinear_interp_np(input_np, out_h, out_w, self.out_size, output_np = bilinear_interp_np(
self.actual_shape, self.align_corners, input_np, out_h, out_w, 0, 0, self.out_size, self.actual_shape,
self.align_mode, self.data_layout) self.align_corners, self.align_mode, self.data_layout)
self.inputs = {'X': input_np} self.inputs = {'X': input_np}
if self.out_size is not None: if self.out_size is not None:
self.inputs['OutSize'] = self.out_size self.inputs['OutSize'] = self.out_size
...@@ -238,6 +247,17 @@ class TestBilinearInterpCase6(TestBilinearInterpOp): ...@@ -238,6 +247,17 @@ class TestBilinearInterpCase6(TestBilinearInterpOp):
self.align_mode = 1 self.align_mode = 1
class TestBilinearInterpCase7(TestBilinearInterpOp):
def init_test_case(self):
self.interp_method = 'bilinear'
self.input_shape = [1, 1, 32, 64]
self.out_h = 64
self.out_w = 32
self.scale = [2.0, 0.5]
self.align_corners = False
self.align_mode = 1
class TestBilinearInterpSame(TestBilinearInterpOp): class TestBilinearInterpSame(TestBilinearInterpOp):
def init_test_case(self): def init_test_case(self):
self.interp_method = 'bilinear' self.interp_method = 'bilinear'
...@@ -298,9 +318,9 @@ class TestBilinearInterpOpUint8(OpTest): ...@@ -298,9 +318,9 @@ class TestBilinearInterpOpUint8(OpTest):
out_h = self.out_h out_h = self.out_h
out_w = self.out_w out_w = self.out_w
output_np = bilinear_interp_np(input_np, out_h, out_w, self.out_size, output_np = bilinear_interp_np(input_np, out_h, out_w, 0, 0,
self.actual_shape, self.align_corners, self.out_size, self.actual_shape,
self.align_mode) self.align_corners, self.align_mode)
self.inputs = {'X': input_np} self.inputs = {'X': input_np}
if self.out_size is not None: if self.out_size is not None:
self.inputs['OutSize'] = self.out_size self.inputs['OutSize'] = self.out_size
...@@ -481,8 +501,9 @@ class TestBilinearInterpOp_attr_tensor(OpTest): ...@@ -481,8 +501,9 @@ class TestBilinearInterpOp_attr_tensor(OpTest):
if isinstance(self.scale, list) and len(self.scale) == 1: if isinstance(self.scale, list) and len(self.scale) == 1:
self.scale = [self.scale[0], self.scale[0]] self.scale = [self.scale[0], self.scale[0]]
self.attrs['scale'] = self.scale self.attrs['scale'] = self.scale
output_np = bilinear_interp_np(input_np, out_h, out_w, self.out_size, output_np = bilinear_interp_np(input_np, out_h, out_w, 0, 0,
self.actual_shape, self.align_corners) self.out_size, self.actual_shape,
self.align_corners)
self.outputs = {'Out': output_np} self.outputs = {'Out': output_np}
def test_check_output(self): def test_check_output(self):
......
python/paddle/fluid/tests/unittests/test_linear_interp_op.py
浏览文件 @ 1cd7d149
...@@ -293,7 +293,7 @@ class TestLinearInterpOpAPI2_0(unittest.TestCase): ...@@ -293,7 +293,7 @@ class TestLinearInterpOpAPI2_0(unittest.TestCase):
# dygraph # dygraph
x_data = np.random.random((1, 3, 128)).astype("float32") x_data = np.random.random((1, 3, 128)).astype("float32")
us_1 = paddle.nn.UpSample( us_1 = paddle.nn.Upsample(
size=[64, ], size=[64, ],
mode='linear', mode='linear',
align_mode=1, align_mode=1,
...@@ -385,19 +385,19 @@ class TestLinearInterpOpError(unittest.TestCase): ...@@ -385,19 +385,19 @@ class TestLinearInterpOpError(unittest.TestCase):
def input_shape_error(): def input_shape_error():
x1 = fluid.data(name="x1", shape=[1], dtype="float32") x1 = fluid.data(name="x1", shape=[1], dtype="float32")
out1 = paddle.nn.UpSample( out1 = paddle.nn.Upsample(
size=[256, ], data_format='NCW', mode='linear') size=[256, ], data_format='NCW', mode='linear')
out1_res = out1(x1) out1_res = out1(x1)
def data_format_error(): def data_format_error():
x2 = fluid.data(name="x2", shape=[1, 3, 128], dtype="float32") x2 = fluid.data(name="x2", shape=[1, 3, 128], dtype="float32")
out2 = paddle.nn.UpSample( out2 = paddle.nn.Upsample(
size=[256, ], data_format='NHWCD', mode='linear') size=[256, ], data_format='NHWCD', mode='linear')
out2_res = out2(x2) out2_res = out2(x2)
def out_shape_error(): def out_shape_error():
x3 = fluid.data(name="x3", shape=[1, 3, 128], dtype="float32") x3 = fluid.data(name="x3", shape=[1, 3, 128], dtype="float32")
out3 = paddle.nn.UpSample( out3 = paddle.nn.Upsample(
size=[ size=[
256, 256,
256, 256,
......
python/paddle/fluid/tests/unittests/test_linear_interp_v2_op.py
浏览文件 @ 1cd7d149
...@@ -26,6 +26,7 @@ from paddle.nn.functional import interpolate ...@@ -26,6 +26,7 @@ from paddle.nn.functional import interpolate
def linear_interp_np(input, def linear_interp_np(input,
out_w, out_w,
scale_w=0,
out_size=None, out_size=None,
actual_shape=None, actual_shape=None,
align_corners=True, align_corners=True,
...@@ -43,6 +44,9 @@ def linear_interp_np(input, ...@@ -43,6 +44,9 @@ def linear_interp_np(input,
if out_w > 1: if out_w > 1:
if (align_corners): if (align_corners):
ratio_w = (in_w - 1.0) / (out_w - 1.0) ratio_w = (in_w - 1.0) / (out_w - 1.0)
else:
if scale_w > 0:
ratio_w = 1.0 / scale_w
else: else:
ratio_w = 1.0 * in_w / out_w ratio_w = 1.0 * in_w / out_w
...@@ -81,6 +85,7 @@ class TestLinearInterpOp(OpTest): ...@@ -81,6 +85,7 @@ class TestLinearInterpOp(OpTest):
self.op_type = "linear_interp_v2" self.op_type = "linear_interp_v2"
input_np = np.random.random(self.input_shape).astype("float64") input_np = np.random.random(self.input_shape).astype("float64")
scale_w = 0
if self.data_layout == "NCHW": if self.data_layout == "NCHW":
in_w = self.input_shape[2] in_w = self.input_shape[2]
else: else:
...@@ -95,7 +100,7 @@ class TestLinearInterpOp(OpTest): ...@@ -95,7 +100,7 @@ class TestLinearInterpOp(OpTest):
else: else:
out_w = self.out_w out_w = self.out_w
output_np = linear_interp_np(input_np, out_w, self.out_size, output_np = linear_interp_np(input_np, out_w, self.scale, self.out_size,
self.actual_shape, self.align_corners, self.actual_shape, self.align_corners,
self.align_mode, self.data_layout) self.align_mode, self.data_layout)
self.inputs = {'X': input_np} self.inputs = {'X': input_np}
...@@ -195,7 +200,7 @@ class TestLinearInterpOpSizeTensor(TestLinearInterpOp): ...@@ -195,7 +200,7 @@ class TestLinearInterpOpSizeTensor(TestLinearInterpOp):
else: else:
out_w = self.out_w out_w = self.out_w
output_np = linear_interp_np(input_np, out_w, self.out_size, output_np = linear_interp_np(input_np, out_w, 0, self.out_size,
self.actual_shape, self.align_corners, self.actual_shape, self.align_corners,
self.align_mode, self.data_layout) self.align_mode, self.data_layout)
...@@ -309,7 +314,7 @@ class TestLinearInterpOpAPI2_0(unittest.TestCase): ...@@ -309,7 +314,7 @@ class TestLinearInterpOpAPI2_0(unittest.TestCase):
# dygraph # dygraph
x_data = np.random.random((1, 3, 128)).astype("float32") x_data = np.random.random((1, 3, 128)).astype("float32")
us_1 = paddle.nn.UpSample( us_1 = paddle.nn.Upsample(
size=[64, ], size=[64, ],
mode='linear', mode='linear',
align_mode=1, align_mode=1,
...@@ -342,7 +347,7 @@ class TestResizeLinearOpUint8(OpTest): ...@@ -342,7 +347,7 @@ class TestResizeLinearOpUint8(OpTest):
else: else:
out_w = self.out_w out_w = self.out_w
output_np = linear_interp_np(input_np, out_w, self.out_size, output_np = linear_interp_np(input_np, out_w, 0, self.out_size,
self.actual_shape, self.align_corners, self.actual_shape, self.align_corners,
self.align_mode) self.align_mode)
self.inputs = {'X': input_np} self.inputs = {'X': input_np}
...@@ -410,19 +415,19 @@ class TestLinearInterpOpError(unittest.TestCase): ...@@ -410,19 +415,19 @@ class TestLinearInterpOpError(unittest.TestCase):
def input_shape_error(): def input_shape_error():
x1 = fluid.data(name="x1", shape=[1], dtype="float32") x1 = fluid.data(name="x1", shape=[1], dtype="float32")
out1 = paddle.nn.UpSample( out1 = paddle.nn.Upsample(
size=[256, ], data_format='NCW', mode='linear') size=[256, ], data_format='NCW', mode='linear')
out1_res = out1(x1) out1_res = out1(x1)
def data_format_error(): def data_format_error():
x2 = fluid.data(name="x2", shape=[1, 3, 128], dtype="float32") x2 = fluid.data(name="x2", shape=[1, 3, 128], dtype="float32")
out2 = paddle.nn.UpSample( out2 = paddle.nn.Upsample(
size=[256, ], data_format='NHWCD', mode='linear') size=[256, ], data_format='NHWCD', mode='linear')
out2_res = out2(x2) out2_res = out2(x2)
def out_shape_error(): def out_shape_error():
x3 = fluid.data(name="x3", shape=[1, 3, 128], dtype="float32") x3 = fluid.data(name="x3", shape=[1, 3, 128], dtype="float32")
out3 = paddle.nn.UpSample( out3 = paddle.nn.Upsample(
size=[ size=[
256, 256,
256, 256,
......
python/paddle/fluid/tests/unittests/test_nearest_interp_v2_op.py
浏览文件 @ 1cd7d149
...@@ -26,6 +26,8 @@ import paddle ...@@ -26,6 +26,8 @@ import paddle
def nearest_neighbor_interp_np(X, def nearest_neighbor_interp_np(X,
out_h, out_h,
out_w, out_w,
scale_h=0,
scale_w=0,
out_size=None, out_size=None,
actual_shape=None, actual_shape=None,
align_corners=True, align_corners=True,
...@@ -45,14 +47,19 @@ def nearest_neighbor_interp_np(X, ...@@ -45,14 +47,19 @@ def nearest_neighbor_interp_np(X,
if (out_h > 1): if (out_h > 1):
if (align_corners): if (align_corners):
ratio_h = (in_h - 1.0) / (out_h - 1.0) ratio_h = (in_h - 1.0) / (out_h - 1.0)
else:
if scale_h > 0:
ratio_h = 1.0 / scale_h
else: else:
ratio_h = 1.0 * in_h / out_h ratio_h = 1.0 * in_h / out_h
if (out_w > 1): if (out_w > 1):
if (align_corners): if (align_corners):
ratio_w = (in_w - 1.0) / (out_w - 1.0) ratio_w = (in_w - 1.0) / (out_w - 1.0)
else:
if scale_w > 0:
ratio_w = 1.0 / scale_w
else: else:
ratio_w = 1.0 * in_w / out_w ratio_w = 1.0 * in_w / out_w
out = np.zeros((n, c, out_h, out_w)) out = np.zeros((n, c, out_h, out_w))
if align_corners: if align_corners:
...@@ -89,7 +96,8 @@ class TestNearestInterpOp(OpTest): ...@@ -89,7 +96,8 @@ class TestNearestInterpOp(OpTest):
else: else:
in_h = self.input_shape[1] in_h = self.input_shape[1]
in_w = self.input_shape[2] in_w = self.input_shape[2]
scale_h = 0
scale_w = 0
if self.scale: if self.scale:
if isinstance(self.scale, float) or isinstance(self.scale, int): if isinstance(self.scale, float) or isinstance(self.scale, int):
if self.scale > 0: if self.scale > 0:
...@@ -106,8 +114,8 @@ class TestNearestInterpOp(OpTest): ...@@ -106,8 +114,8 @@ class TestNearestInterpOp(OpTest):
out_w = self.out_w out_w = self.out_w
output_np = nearest_neighbor_interp_np( output_np = nearest_neighbor_interp_np(
input_np, out_h, out_w, self.out_size, self.actual_shape, input_np, out_h, out_w, scale_h, scale_w, self.out_size,
self.align_corners, self.data_layout) self.actual_shape, self.align_corners, self.data_layout)
self.inputs = {'X': input_np} self.inputs = {'X': input_np}
if self.out_size is not None: if self.out_size is not None:
self.inputs['OutSize'] = self.out_size self.inputs['OutSize'] = self.out_size
...@@ -265,7 +273,7 @@ class TestNearestInterpOpUint8(OpTest): ...@@ -265,7 +273,7 @@ class TestNearestInterpOpUint8(OpTest):
out_h = self.out_h out_h = self.out_h
out_w = self.out_w out_w = self.out_w
output_np = nearest_neighbor_interp_np(input_np, out_h, out_w, output_np = nearest_neighbor_interp_np(input_np, out_h, out_w, 0, 0,
self.out_size, self.actual_shape, self.out_size, self.actual_shape,
self.align_corners) self.align_corners)
self.inputs = {'X': input_np} self.inputs = {'X': input_np}
...@@ -408,7 +416,7 @@ class TestNearestInterpOp_attr_tensor(OpTest): ...@@ -408,7 +416,7 @@ class TestNearestInterpOp_attr_tensor(OpTest):
if isinstance(self.scale, list) and len(self.scale) == 1: if isinstance(self.scale, list) and len(self.scale) == 1:
self.scale = [self.scale[0], self.scale[0]] self.scale = [self.scale[0], self.scale[0]]
self.attrs['scale'] = self.scale self.attrs['scale'] = self.scale
output_np = nearest_neighbor_interp_np(input_np, out_h, out_w, output_np = nearest_neighbor_interp_np(input_np, out_h, out_w, 0, 0,
self.out_size, self.actual_shape, self.out_size, self.actual_shape,
self.align_corners) self.align_corners)
self.outputs = {'Out': output_np} self.outputs = {'Out': output_np}
......
python/paddle/fluid/tests/unittests/test_trilinear_interp_v2_op.py
浏览文件 @ 1cd7d149
...@@ -26,6 +26,9 @@ def trilinear_interp_np(input, ...@@ -26,6 +26,9 @@ def trilinear_interp_np(input,
out_d, out_d,
out_h, out_h,
out_w, out_w,
scale_d=0,
scale_h=0,
scale_w=0,
out_size=None, out_size=None,
actual_shape=None, actual_shape=None,
align_corners=True, align_corners=True,
...@@ -48,16 +51,25 @@ def trilinear_interp_np(input, ...@@ -48,16 +51,25 @@ def trilinear_interp_np(input,
if out_d > 1: if out_d > 1:
if (align_corners): if (align_corners):
ratio_d = (in_d - 1.0) / (out_d - 1.0) ratio_d = (in_d - 1.0) / (out_d - 1.0)
else:
if scale_d > 0:
ratio_d = 1.0 / scale_d
else: else:
ratio_d = 1.0 * in_d / out_d ratio_d = 1.0 * in_d / out_d
if out_h > 1: if out_h > 1:
if (align_corners): if (align_corners):
ratio_h = (in_h - 1.0) / (out_h - 1.0) ratio_h = (in_h - 1.0) / (out_h - 1.0)
else:
if scale_h > 0:
ratio_h = 1.0 / scale_h
else: else:
ratio_h = 1.0 * in_h / out_h ratio_h = 1.0 * in_h / out_h
if out_w > 1: if out_w > 1:
if (align_corners): if (align_corners):
ratio_w = (in_w - 1.0) / (out_w - 1.0) ratio_w = (in_w - 1.0) / (out_w - 1.0)
else:
if scale_w > 0:
ratio_w = 1.0 / scale_w
else: else:
ratio_w = 1.0 * in_w / out_w ratio_w = 1.0 * in_w / out_w
...@@ -133,6 +145,9 @@ class TestTrilinearInterpOp(OpTest): ...@@ -133,6 +145,9 @@ class TestTrilinearInterpOp(OpTest):
self.op_type = "trilinear_interp_v2" self.op_type = "trilinear_interp_v2"
input_np = np.random.random(self.input_shape).astype("float32") input_np = np.random.random(self.input_shape).astype("float32")
scale_w = 0
scale_h = 0
scale_d = 0
if self.data_layout == "NCDHW": if self.data_layout == "NCDHW":
in_d = self.input_shape[2] in_d = self.input_shape[2]
in_h = self.input_shape[3] in_h = self.input_shape[3]
...@@ -159,9 +174,10 @@ class TestTrilinearInterpOp(OpTest): ...@@ -159,9 +174,10 @@ class TestTrilinearInterpOp(OpTest):
out_h = self.out_h out_h = self.out_h
out_w = self.out_w out_w = self.out_w
output_np = trilinear_interp_np( output_np = trilinear_interp_np(input_np, out_d, out_h, out_w, scale_d,
input_np, out_d, out_h, out_w, self.out_size, self.actual_shape, scale_h, scale_w, self.out_size,
self.align_corners, self.align_mode, self.data_layout) self.actual_shape, self.align_corners,
self.align_mode, self.data_layout)
self.inputs = {'X': input_np} self.inputs = {'X': input_np}
if self.out_size is not None: if self.out_size is not None:
self.inputs['OutSize'] = self.out_size self.inputs['OutSize'] = self.out_size
...@@ -359,7 +375,7 @@ class TestTrilinearInterpOpUint8(OpTest): ...@@ -359,7 +375,7 @@ class TestTrilinearInterpOpUint8(OpTest):
out_h = self.out_h out_h = self.out_h
out_w = self.out_w out_w = self.out_w
output_np = trilinear_interp_np(input_np, out_d, out_h, out_w, output_np = trilinear_interp_np(input_np, out_d, out_h, out_w, 0, 0, 0,
self.out_size, self.actual_shape, self.out_size, self.actual_shape,
self.align_corners, self.align_mode) self.align_corners, self.align_mode)
self.inputs = {'X': input_np} self.inputs = {'X': input_np}
...@@ -482,7 +498,7 @@ class TestTrilinearInterpZero(TestTrilinearInterpOp): ...@@ -482,7 +498,7 @@ class TestTrilinearInterpZero(TestTrilinearInterpOp):
self.out_d = 60 self.out_d = 60
self.out_h = 40 self.out_h = 40
self.out_w = 25 self.out_w = 25
self.scale = 0.2 self.scale = 0.0
self.align_corners = False self.align_corners = False
self.align_mode = 0 self.align_mode = 0
...@@ -541,7 +557,7 @@ class TestTrilinearInterpOp_attr_tensor(OpTest): ...@@ -541,7 +557,7 @@ class TestTrilinearInterpOp_attr_tensor(OpTest):
if isinstance(self.scale, list) and len(self.scale) == 1: if isinstance(self.scale, list) and len(self.scale) == 1:
self.scale = [self.scale[0], self.scale[0], self.scale[0]] self.scale = [self.scale[0], self.scale[0], self.scale[0]]
self.attrs['scale'] = self.scale self.attrs['scale'] = self.scale
output_np = trilinear_interp_np(input_np, out_d, out_h, out_w, output_np = trilinear_interp_np(input_np, out_d, out_h, out_w, 0, 0, 0,
self.out_size, self.actual_shape, self.out_size, self.actual_shape,
self.align_corners, self.align_mode) self.align_corners, self.align_mode)
self.outputs = {'Out': output_np} self.outputs = {'Out': output_np}
......
python/paddle/nn/__init__.py
浏览文件 @ 1cd7d149
...@@ -89,7 +89,7 @@ from .layer.common import CosineSimilarity #DEFINE_ALIAS ...@@ -89,7 +89,7 @@ from .layer.common import CosineSimilarity #DEFINE_ALIAS
from .layer.common import Embedding #DEFINE_ALIAS from .layer.common import Embedding #DEFINE_ALIAS
from .layer.common import Linear #DEFINE_ALIAS from .layer.common import Linear #DEFINE_ALIAS
from .layer.common import Flatten #DEFINE_ALIAS from .layer.common import Flatten #DEFINE_ALIAS
from .layer.common import UpSample #DEFINE_ALIAS from .layer.common import Upsample #DEFINE_ALIAS
from .layer.common import UpsamplingNearest2d #DEFINE_ALIAS from .layer.common import UpsamplingNearest2d #DEFINE_ALIAS
from .layer.common import UpsamplingBilinear2d #DEFINE_ALIAS from .layer.common import UpsamplingBilinear2d #DEFINE_ALIAS
from .layer.common import Bilinear #DEFINE_ALIAS from .layer.common import Bilinear #DEFINE_ALIAS
......
python/paddle/nn/functional/__init__.py
浏览文件 @ 1cd7d149
...@@ -72,6 +72,7 @@ from .common import unfold #DEFINE_ALIAS ...@@ -72,6 +72,7 @@ from .common import unfold #DEFINE_ALIAS
# from .common import bilinear_tensor_product #DEFINE_ALIAS # from .common import bilinear_tensor_product #DEFINE_ALIAS
from .common import assign #DEFINE_ALIAS from .common import assign #DEFINE_ALIAS
from .common import interpolate #DEFINE_ALIAS from .common import interpolate #DEFINE_ALIAS
from .common import upsample #DEFINE_ALIAS
from .common import bilinear #DEFINE_ALIAS from .common import bilinear #DEFINE_ALIAS
from .conv import conv1d #DEFINE_ALIAS from .conv import conv1d #DEFINE_ALIAS
from .conv import conv_transpose1d #DEFINE_ALIAS from .conv import conv_transpose1d #DEFINE_ALIAS
......
python/paddle/nn/functional/common.py
浏览文件 @ 1cd7d149
...@@ -80,6 +80,8 @@ def interpolate(x, ...@@ -80,6 +80,8 @@ def interpolate(x,
The input must be a 3-D Tensor of the shape (num_batches, channels, in_w) The input must be a 3-D Tensor of the shape (num_batches, channels, in_w)
or 4-D (num_batches, channels, in_h, in_w), or a 5-D Tensor of the shape or 4-D (num_batches, channels, in_h, in_w), or a 5-D Tensor of the shape
(num_batches, channels, in_d, in_h, in_w) or (num_batches, in_d, in_h, in_w, channels), (num_batches, channels, in_d, in_h, in_w) or (num_batches, in_d, in_h, in_w, channels),
Where in_w is width of the input tensor, in_h is the height of the input tensor,
in_d is the depth of the intput tensor.
and the resizing only applies on the three dimensions(depth, height and width). and the resizing only applies on the three dimensions(depth, height and width).
Supporting resample methods: Supporting resample methods:
...@@ -88,6 +90,7 @@ def interpolate(x, ...@@ -88,6 +90,7 @@ def interpolate(x,
'trilinear' : Trilinear interpolation 'trilinear' : Trilinear interpolation
'nearest' : Nearest neighbor interpolation 'nearest' : Nearest neighbor interpolation
'bicubic' : Bicubic interpolation 'bicubic' : Bicubic interpolation
'area': Area interpolation
Linear interpolation is the method of using a line connecting two known quantities Linear interpolation is the method of using a line connecting two known quantities
to determine the value of an unknown quantity between the two known quantities. to determine the value of an unknown quantity between the two known quantities.
...@@ -114,6 +117,12 @@ def interpolate(x, ...@@ -114,6 +117,12 @@ def interpolate(x,
smoother than corresponding surfaces obtained by bilinear interpolation or smoother than corresponding surfaces obtained by bilinear interpolation or
nearest-neighbor interpolation. nearest-neighbor interpolation.
Area interpolation is to perform area interpolation
in both the 3rd dimension(in height direction) , the 4th dimension(in width
direction) and the 5th dimension(in depth direction) on input tensor. Set to
area will directly call `paddle.nn.functional.adaptive_avg_pool1d` or
`paddle.nn.functional.adaptive_avg_pool2d` or `paddle.nn.functional.adaptive_avg_pool3d`.
Example: Example:
.. code-block:: text .. code-block:: text
...@@ -207,11 +216,11 @@ def interpolate(x, ...@@ -207,11 +216,11 @@ def interpolate(x,
when input is a 4-D Tensor and is (out_d, out_h, out_w) when input is a 5-D Tensor. when input is a 4-D Tensor and is (out_d, out_h, out_w) when input is a 5-D Tensor.
Default: None. If a list, each element can be an integer or a Tensor Variable of shape: [1]. Default: None. If a list, each element can be an integer or a Tensor Variable of shape: [1].
If a Tensor Variable, its dimensions size should be a 1. If a Tensor Variable, its dimensions size should be a 1.
scale_factor (float|Tensor|list|None): The multiplier for the input height or width. At scale_factor (float|Tensor|list|tuple|None): The multiplier for the input height or width. At
least one of :attr:`out_shape` or :attr:`scale_factor` must be set. least one of :attr:`size` or :attr:`scale_factor` must be set.
And :attr:`out_shape` has a higher priority than :attr:`scale_factor`.Has to match input size if it is a list. And :attr:`size` has a higher priority than :attr:`scale_factor`.Has to match input size if it is either a list or a tuple or a Tensor.
Default: None. Default: None.
mode (str): The resample method. It supports 'linear', 'nearest', 'bilinear', mode (str): The resample method. It supports 'linear', 'area', 'nearest', 'bilinear',
'bicubic' and 'trilinear' currently. Default: 'nearest' 'bicubic' and 'trilinear' currently. Default: 'nearest'
align_corners(bool) : An optional bool, If True, the centers of the 4 corner pixels of the align_corners(bool) : An optional bool, If True, the centers of the 4 corner pixels of the
input and output tensors are aligned, preserving the values at the input and output tensors are aligned, preserving the values at the
...@@ -235,7 +244,7 @@ def interpolate(x, ...@@ -235,7 +244,7 @@ def interpolate(x,
Raises: Raises:
TypeError: size should be a list or tuple or Tensor. TypeError: size should be a list or tuple or Tensor.
ValueError: The 'mode' of image_resize can only be 'linear', 'bilinear', ValueError: The 'mode' of image_resize can only be 'linear', 'bilinear',
'trilinear', 'bicubic', or 'nearest' currently. 'trilinear', 'bicubic', 'area' or 'nearest' currently.
ValueError: 'linear' only support 3-D tensor. ValueError: 'linear' only support 3-D tensor.
ValueError: 'bilinear', 'bicubic' and 'nearest' only support 4-D tensor. ValueError: 'bilinear', 'bicubic' and 'nearest' only support 4-D tensor.
ValueError: 'trilinear' only support 5-D tensor. ValueError: 'trilinear' only support 5-D tensor.
...@@ -283,10 +292,11 @@ def interpolate(x, ...@@ -283,10 +292,11 @@ def interpolate(x,
'TRILINEAR', 'TRILINEAR',
'NEAREST', 'NEAREST',
'BICUBIC', 'BICUBIC',
'AREA',
] ]
if resample not in resample_methods: if resample not in resample_methods:
raise ValueError( raise ValueError(
"The 'resample' of image_resize can only be 'linaer', 'bilinear', 'trilinear', " "The 'resample' of image_resize can only be 'area', 'linear', 'bilinear', 'trilinear', "
" 'bicubic' or 'nearest' currently.") " 'bicubic' or 'nearest' currently.")
if resample in ['LINEAR'] and len(x.shape) != 3: if resample in ['LINEAR'] and len(x.shape) != 3:
...@@ -310,8 +320,17 @@ def interpolate(x, ...@@ -310,8 +320,17 @@ def interpolate(x,
raise ValueError( raise ValueError(
"align_corners option can only be set with the interpolating modes: linear | bilinear | bicubic | trilinear" "align_corners option can only be set with the interpolating modes: linear | bilinear | bicubic | trilinear"
) )
if resample == 'AREA' and len(x.shape) == 3:
return paddle.nn.functional.adaptive_avg_pool1d(x, size)
if resample == 'AREA' and len(x.shape) == 4:
return paddle.nn.functional.adaptive_avg_pool2d(x, size)
if resample == 'AREA' and len(x.shape) == 5:
return paddle.nn.functional.adaptive_avg_pool3d(x, size)
helper = LayerHelper('{}_interp_v2'.format(resample_type), **locals()) helper = LayerHelper('{}_interp_v2'.format(resample_type), **locals())
dtype = helper.input_dtype() dtype = helper.input_dtype(input_param_name='x')
if len(x.shape) == 3 and data_format not in ['NCW', 'NWC']: if len(x.shape) == 3 and data_format not in ['NCW', 'NWC']:
raise ValueError( raise ValueError(
"Got wrong value for param `data_format`: " + data_format + "Got wrong value for param `data_format`: " + data_format +
...@@ -349,14 +368,15 @@ def interpolate(x, ...@@ -349,14 +368,15 @@ def interpolate(x,
out_shape = size out_shape = size
scale = scale_factor scale = scale_factor
if out_shape is not None and scale is not None:
raise ValueError("Only one of size or scale_factor should be defined.")
if out_shape is not None: if out_shape is not None:
if isinstance(out_shape, Variable): if isinstance(out_shape, Variable):
out_shape.stop_gradient = True out_shape.stop_gradient = True
inputs['OutSize'] = out_shape inputs['OutSize'] = out_shape
else: else:
if not (_is_list_or_turple_(out_shape)): if not (_is_list_or_turple_(out_shape)):
raise TypeError( raise TypeError("size should be a list or tuple or Variable.")
"out_shape should be a list or tuple or Variable.")
# Validate the shape # Validate the shape
contain_var = False contain_var = False
for dim_idx, dim_size in enumerate(out_shape): for dim_idx, dim_size in enumerate(out_shape):
...@@ -388,7 +408,7 @@ def interpolate(x, ...@@ -388,7 +408,7 @@ def interpolate(x,
if len(x.shape) == 3: if len(x.shape) == 3:
if len(out_shape) != 1: if len(out_shape) != 1:
raise ValueError( raise ValueError(
"out_shape length should be 2 for input 3-D tensor") "size length should be 2 for input 3-D tensor")
if contain_var: if contain_var:
attrs['out_w'] = size_list[0] attrs['out_w'] = size_list[0]
else: else:
...@@ -396,7 +416,7 @@ def interpolate(x, ...@@ -396,7 +416,7 @@ def interpolate(x,
attrs['out_w'] = out_shape[0] attrs['out_w'] = out_shape[0]
if len(x.shape) == 4: if len(x.shape) == 4:
if len(out_shape) != 2: if len(out_shape) != 2:
raise ValueError("out_shape length should be 2 for " raise ValueError("size length should be 2 for "
"input 4-D tensor.") "input 4-D tensor.")
if contain_var: if contain_var:
attrs['out_h'] = size_list[0] attrs['out_h'] = size_list[0]
...@@ -407,7 +427,7 @@ def interpolate(x, ...@@ -407,7 +427,7 @@ def interpolate(x,
attrs['out_w'] = out_shape[1] attrs['out_w'] = out_shape[1]
if len(x.shape) == 5: if len(x.shape) == 5:
if len(out_shape) != 3: if len(out_shape) != 3:
raise ValueError("out_shape length should be 3 for " raise ValueError("size length should be 3 for "
"input 5-D tensor.") "input 5-D tensor.")
if contain_var: if contain_var:
attrs['out_d'] = size_list[0] attrs['out_d'] = size_list[0]
...@@ -430,7 +450,7 @@ def interpolate(x, ...@@ -430,7 +450,7 @@ def interpolate(x,
for i in range(len(x.shape) - 2): for i in range(len(x.shape) - 2):
scale_list.append(scale) scale_list.append(scale)
attrs['scale'] = list(map(float, scale_list)) attrs['scale'] = list(map(float, scale_list))
elif isinstance(scale, list): elif isinstance(scale, list) or isinstance(scale, float):
if len(scale) != len(x.shape) - 2: if len(scale) != len(x.shape) - 2:
raise ValueError("scale_shape length should be {} for " raise ValueError("scale_shape length should be {} for "
"input {}-D tensor.".format( "input {}-D tensor.".format(
...@@ -441,7 +461,8 @@ def interpolate(x, ...@@ -441,7 +461,8 @@ def interpolate(x,
attrs['scale'] = list(map(float, scale)) attrs['scale'] = list(map(float, scale))
else: else:
raise TypeError( raise TypeError(
"Attr(scale)'s type should be float, int, list or Tensor.") "Attr(scale)'s type should be float, int, list, tuple, or Tensor."
)
if in_dygraph_mode(): if in_dygraph_mode():
attr_list = [] attr_list = []
...@@ -480,9 +501,12 @@ def upsample(x, ...@@ -480,9 +501,12 @@ def upsample(x,
name=None): name=None):
""" """
This op resizes a batch of images. This op resizes a batch of images.
The input must be a 3-D Tensor of the shape (num_batches, channels, in_w) The input must be a 3-D Tensor of the shape (num_batches, channels, in_w)
or 4-D (num_batches, channels, in_h, in_w), or a 5-D Tensor of the shape or 4-D (num_batches, channels, in_h, in_w), or a 5-D Tensor of the shape
(num_batches, channels, in_d, in_h, in_w) or (num_batches, in_d, in_h, in_w, channels), (num_batches, channels, in_d, in_h, in_w) or (num_batches, in_d, in_h, in_w, channels),
Where in_w is width of the input tensor, in_h is the height of the input tensor,
in_d is the depth of the intput tensor.
and the resizing only applies on the three dimensions(depth, height and width). and the resizing only applies on the three dimensions(depth, height and width).
Supporting resample methods: Supporting resample methods:
...@@ -507,12 +531,21 @@ def upsample(x, ...@@ -507,12 +531,21 @@ def upsample(x,
data points on a two-dimensional regular grid. The interpolated surface is data points on a two-dimensional regular grid. The interpolated surface is
smoother than corresponding surfaces obtained by bilinear interpolation or smoother than corresponding surfaces obtained by bilinear interpolation or
nearest-neighbor interpolation. nearest-neighbor interpolation.
Trilinear interpolation is an extension of linear interpolation for Trilinear interpolation is an extension of linear interpolation for
interpolating functions of three variables (e.g. D-direction, interpolating functions of three variables (e.g. D-direction,
H-direction and W-direction in this op) on a rectilinear 3D grid. H-direction and W-direction in this op) on a rectilinear 3D grid.
The linear interpolation is performed on three directions. The linear interpolation is performed on three directions.
align_corners and align_mode are optional parameters,the calculation method align_corners and align_mode are optional parameters,the calculation method
of interpolation can be selected by them. of interpolation can be selected by them.
Area interpolation is to perform area interpolation
in both the 3rd dimension(in height direction) , the 4th dimension(in width
direction) and the 5th dimension(in depth direction) on input tensor. Set to
area will directly call `paddle.nn.functional.adaptive_avg_pool1d` or
`paddle.nn.functional.adaptive_avg_pool2d` or `paddle.nn.functional.adaptive_avg_pool3d`.
Example: Example:
.. code-block:: text .. code-block:: text
For scale_factor: For scale_factor:
...@@ -605,9 +638,10 @@ def upsample(x, ...@@ -605,9 +638,10 @@ def upsample(x,
when input is a 4-D Tensor and is (out_d, out_h, out_w) when input is a 5-D Tensor. when input is a 4-D Tensor and is (out_d, out_h, out_w) when input is a 5-D Tensor.
Default: None. If a list, each element can be an integer or a Tensor Variable of shape: [1]. Default: None. If a list, each element can be an integer or a Tensor Variable of shape: [1].
If a Tensor Variable, its dimensions size should be a 1. If a Tensor Variable, its dimensions size should be a 1.
scale_factor (float|Tensor|list|None): The multiplier for the input height or width. At scale_factor (float|Tensor|list|tuple|None): The multiplier for the input height or width. At
least one of :attr:`out_shape` or :attr:`scale_factor` must be set. least one of :attr:`size` or :attr:`scale_factor` must be set.
And :attr:`out_shape` has a higher priority than :attr:`scale_factor`. And :attr:`size` has a higher priority than :attr:`scale_factor`.Has to match input size if
it is either a list or a tuple or a Tensor.
Default: None. Default: None.
mode (str): The resample method. It supports 'linear', 'nearest', 'bilinear', mode (str): The resample method. It supports 'linear', 'nearest', 'bilinear',
'bicubic' and 'trilinear' currently. Default: 'nearest' 'bicubic' and 'trilinear' currently. Default: 'nearest'
......
python/paddle/nn/layer/__init__.py
浏览文件 @ 1cd7d149
...@@ -59,7 +59,7 @@ from .common import CosineSimilarity #DEFINE_ALIAS ...@@ -59,7 +59,7 @@ from .common import CosineSimilarity #DEFINE_ALIAS
from .common import Embedding #DEFINE_ALIAS from .common import Embedding #DEFINE_ALIAS
from .common import Linear #DEFINE_ALIAS from .common import Linear #DEFINE_ALIAS
from .common import Flatten #DEFINE_ALIAS from .common import Flatten #DEFINE_ALIAS
from .common import UpSample #DEFINE_ALIAS from .common import Upsample #DEFINE_ALIAS
from .common import UpsamplingNearest2d #DEFINE_ALIAS from .common import UpsamplingNearest2d #DEFINE_ALIAS
from .common import UpsamplingBilinear2d #DEFINE_ALIAS from .common import UpsamplingBilinear2d #DEFINE_ALIAS
from .common import Dropout #DEFINE_ALIAS from .common import Dropout #DEFINE_ALIAS
......
python/paddle/nn/layer/common.py
浏览文件 @ 1cd7d149
...@@ -26,7 +26,7 @@ __all__ = [ ...@@ -26,7 +26,7 @@ __all__ = [
'Pool2D', 'Pool2D',
'Embedding', 'Embedding',
'Linear', 'Linear',
'UpSample', 'Upsample',
'Pad2D', 'Pad2D',
'UpsamplingNearest2d', 'UpsamplingNearest2d',
'UpsamplingBilinear2d', 'UpsamplingBilinear2d',
...@@ -131,12 +131,15 @@ class Linear(layers.Layer): ...@@ -131,12 +131,15 @@ class Linear(layers.Layer):
return out return out
class UpSample(layers.Layer): class Upsample(layers.Layer):
""" """
This op resizes a batch of images. This op resizes a batch of images.
The input must be a 3-D Tensor of the shape (num_batches, channels, in_w) The input must be a 3-D Tensor of the shape (num_batches, channels, in_w)
or 4-D (num_batches, channels, in_h, in_w), or a 5-D Tensor of the shape or 4-D (num_batches, channels, in_h, in_w), or a 5-D Tensor of the shape
(num_batches, channels, in_d, in_h, in_w) or (num_batches, in_d, in_h, in_w, channels), (num_batches, channels, in_d, in_h, in_w) or (num_batches, in_d, in_h, in_w, channels),
Where in_w is width of the input tensor, in_h is the height of the input tensor,
in_d is the depth of the intput tensor.
and the resizing only applies on the three dimensions(depth, height and width). and the resizing only applies on the three dimensions(depth, height and width).
Supporting resample methods: Supporting resample methods:
...@@ -171,6 +174,12 @@ class UpSample(layers.Layer): ...@@ -171,6 +174,12 @@ class UpSample(layers.Layer):
align_corners and align_mode are optional parameters,the calculation method align_corners and align_mode are optional parameters,the calculation method
of interpolation can be selected by them. of interpolation can be selected by them.
Area interpolation is to perform area interpolation
in both the 3rd dimension(in height direction) , the 4th dimension(in width
direction) and the 5th dimension(in depth direction) on input tensor. Set to
area will directly call `paddle.nn.functional.adaptive_avg_pool1d` or
`paddle.nn.functional.adaptive_avg_pool2d` or `paddle.nn.functional.adaptive_avg_pool3d`.
Example: Example:
.. code-block:: text .. code-block:: text
...@@ -273,9 +282,9 @@ class UpSample(layers.Layer): ...@@ -273,9 +282,9 @@ class UpSample(layers.Layer):
when input is a 4-D Tensor and is (out_d, out_h, out_w) when input is a 5-D Tensor. when input is a 4-D Tensor and is (out_d, out_h, out_w) when input is a 5-D Tensor.
Default: None. If a list, each element can be an integer or a Tensor Variable of shape: [1]. Default: None. If a list, each element can be an integer or a Tensor Variable of shape: [1].
If a Tensor Variable, its dimensions size should be a 1. If a Tensor Variable, its dimensions size should be a 1.
scale_factor (float|Tensor|list|None): The multiplier for the input height or width. At scale_factor (float|Tensor|list|tuple|None): The multiplier for the input height or width. At
least one of :attr:`out_shape` or :attr:`scale_factor` must be set. least one of :attr:`size` or :attr:`scale_factor` must be set.
And :attr:`out_shape` has a higher priority than :attr:`scale_factor`.Has to match input size if it is a list. And :attr:`size` has a higher priority than :attr:`scale_factor`. Has to match input size if it is either a list or a tuple or a Tensor.
Default: None. Default: None.
mode (str): The resample method. It supports 'linear', 'nearst', 'bilinear', mode (str): The resample method. It supports 'linear', 'nearst', 'bilinear',
'bicubic' and 'trilinear' currently. Default: 'nearest' 'bicubic' and 'trilinear' currently. Default: 'nearest'
...@@ -322,7 +331,7 @@ class UpSample(layers.Layer): ...@@ -322,7 +331,7 @@ class UpSample(layers.Layer):
paddle.disable_static() paddle.disable_static()
input_data = np.random.rand(2,3,6,10).astype("float32") input_data = np.random.rand(2,3,6,10).astype("float32")
upsample_out = paddle.nn.UpSample(size=[12,12]) upsample_out = paddle.nn.Upsample(size=[12,12])
input = paddle.to_tensor(input_data) input = paddle.to_tensor(input_data)
output = upsample_out(x=input) output = upsample_out(x=input)
...@@ -339,7 +348,7 @@ class UpSample(layers.Layer): ...@@ -339,7 +348,7 @@ class UpSample(layers.Layer):
align_mode=0, align_mode=0,
data_format='NCHW', data_format='NCHW',
name=None): name=None):
super(UpSample, self).__init__() super(Upsample, self).__init__()
self.size = size self.size = size
self.scale_factor = scale_factor self.scale_factor = scale_factor
self.mode = mode.lower() self.mode = mode.lower()
...@@ -366,7 +375,8 @@ class UpsamplingNearest2d(layers.Layer): ...@@ -366,7 +375,8 @@ class UpsamplingNearest2d(layers.Layer):
""" """
This op upsamples a batch of images, using nearest neighbours' pixel values. This op upsamples a batch of images, using nearest neighbours' pixel values.
The input must be a 4-D Tensor of the shape (num_batches, channels, in_h, in_w), The input must be a 4-D Tensor of the shape (num_batches, channels, in_h, in_w),
and the upsampling only applies on the two dimensions(height and width). where in_w is width of the input tensor, in_h is the height of the input tensor.
And the upsampling only applies on the two dimensions(height and width).
Nearest neighbor interpolation is to perform nearest neighbor interpolation Nearest neighbor interpolation is to perform nearest neighbor interpolation
in both the 3rd dimension(in height direction) and the 4th dimension(in width in both the 3rd dimension(in height direction) and the 4th dimension(in width
...@@ -381,10 +391,11 @@ class UpsamplingNearest2d(layers.Layer): ...@@ -381,10 +391,11 @@ class UpsamplingNearest2d(layers.Layer):
layer, the shape is (out_h, out_w) when input is a 4-D Tensor. layer, the shape is (out_h, out_w) when input is a 4-D Tensor.
Default: None. If a list, each element can be an integer or a Tensor Variable of shape: [1]. Default: None. If a list, each element can be an integer or a Tensor Variable of shape: [1].
If a Tensor Variable, its dimensions size should be a 1. If a Tensor Variable, its dimensions size should be a 1.
scale_factor (float|int|list|Tensor|None): The multiplier for the input height or width. At scale_factor (float|int|list|tuple|Tensor|None): The multiplier for the input height or width. At
least one of :attr:`out_shape` or :attr:`scale_factor` must be set. least one of :attr:`size` or :attr:`scale_factor` must be set.
And :attr:`out_shape` has a higher priority than :attr:`scale_factor`. And :attr:`size` has a higher priority than :attr:`scale_factor`.
Default: None. Has to match input size if it is a list. Has to match input size if it is either a list or a tuple or a Tensor.
Default: None.
data_format (str, optional): Specify the data format of the input, and the data format of the output data_format (str, optional): Specify the data format of the input, and the data format of the output
will be consistent with that of the input. An optional string from:`NCW`, `NWC`, `"NCHW"`, `"NHWC"`, `"NCDHW"`, will be consistent with that of the input. An optional string from:`NCW`, `NWC`, `"NCHW"`, `"NHWC"`, `"NCDHW"`,
`"NDHWC"`. The default is `"NCHW"`. When it is `"NCHW"`, the data is stored in the order of: `"NDHWC"`. The default is `"NCHW"`. When it is `"NCHW"`, the data is stored in the order of:
...@@ -449,7 +460,8 @@ class UpsamplingBilinear2d(layers.Layer): ...@@ -449,7 +460,8 @@ class UpsamplingBilinear2d(layers.Layer):
""" """
This op upsamples a batch of images, using bilinear' pixel values. This op upsamples a batch of images, using bilinear' pixel values.
The input must be a 4-D Tensor of the shape (num_batches, channels, in_h, in_w), The input must be a 4-D Tensor of the shape (num_batches, channels, in_h, in_w),
and the upsampling only applies on the two dimensions(height and width). where in_w is width of the input tensor, in_h is the height of the input tensor.
And the upsampling only applies on the two dimensions(height and width).
Bilinear interpolation is an extension of linear interpolation for Bilinear interpolation is an extension of linear interpolation for
interpolating functions of two variables (e.g. H-direction and interpolating functions of two variables (e.g. H-direction and
...@@ -466,10 +478,11 @@ class UpsamplingBilinear2d(layers.Layer): ...@@ -466,10 +478,11 @@ class UpsamplingBilinear2d(layers.Layer):
layer, the shape is (out_h, out_w) when input is a 4-D Tensor. layer, the shape is (out_h, out_w) when input is a 4-D Tensor.
Default: None. If a list, each element can be an integer or a Tensor Variable of shape: [1]. Default: None. If a list, each element can be an integer or a Tensor Variable of shape: [1].
If a Tensor Variable, its dimensions size should be a 1. If a Tensor Variable, its dimensions size should be a 1.
scale_factor (float|int|list|Tensor|None): The multiplier for the input height or width. At scale_factor (float|int|list|tuple|Tensor|None): The multiplier for the input height or width. At
least one of :attr:`out_shape` or :attr:`scale_factor` must be set. least one of :attr:`size` or :attr:`scale_factor` must be set.
And :attr:`out_shape` has a higher priority than :attr:`scale_factor`. And :attr:`size` has a higher priority than :attr:`scale_factor`.
Default: None. Has to match input size if it is a list. Has to match input size if it is either a list or a tuple or a Tensor.
Default: None.
data_format (str, optional): Specify the data format of the input, and the data format of the output data_format (str, optional): Specify the data format of the input, and the data format of the output
will be consistent with that of the input. An optional string from:`NCW`, `NWC`, `"NCHW"`, `"NHWC"`, `"NCDHW"`, will be consistent with that of the input. An optional string from:`NCW`, `NWC`, `"NCHW"`, `"NHWC"`, `"NCDHW"`,
`"NDHWC"`. The default is `"NCHW"`. When it is `"NCHW"`, the data is stored in the order of: `"NDHWC"`. The default is `"NCHW"`. When it is `"NCHW"`, the data is stored in the order of:
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或