提交 f06c6193 编写于 作者: J jerrywgz

fix rpn target assign test=develop

上级 765085d2
...@@ -52,6 +52,9 @@ class RpnTargetAssignOp : public framework::OperatorWithKernel { ...@@ -52,6 +52,9 @@ class RpnTargetAssignOp : public framework::OperatorWithKernel {
PADDLE_ENFORCE( PADDLE_ENFORCE(
ctx->HasOutput("TargetBBox"), ctx->HasOutput("TargetBBox"),
"Output(TargetBBox) of RpnTargetAssignOp should not be null"); "Output(TargetBBox) of RpnTargetAssignOp should not be null");
PADDLE_ENFORCE(
ctx->HasOutput("BBox_inside_weight"),
"Output(BBox_inside_weight) of RpnTargetAssignOp should not be null");
auto anchor_dims = ctx->GetInputDim("Anchor"); auto anchor_dims = ctx->GetInputDim("Anchor");
auto gt_boxes_dims = ctx->GetInputDim("GtBoxes"); auto gt_boxes_dims = ctx->GetInputDim("GtBoxes");
...@@ -68,6 +71,7 @@ class RpnTargetAssignOp : public framework::OperatorWithKernel { ...@@ -68,6 +71,7 @@ class RpnTargetAssignOp : public framework::OperatorWithKernel {
ctx->SetOutputDim("ScoreIndex", {-1}); ctx->SetOutputDim("ScoreIndex", {-1});
ctx->SetOutputDim("TargetLabel", {-1, 1}); ctx->SetOutputDim("TargetLabel", {-1, 1});
ctx->SetOutputDim("TargetBBox", {-1, 4}); ctx->SetOutputDim("TargetBBox", {-1, 4});
ctx->SetOutputDim("BBox_inside_weight", {-1, 4});
} }
protected: protected:
...@@ -169,6 +173,7 @@ void ScoreAssign(const T* anchor_by_gt_overlap_data, ...@@ -169,6 +173,7 @@ void ScoreAssign(const T* anchor_by_gt_overlap_data,
const float rpn_positive_overlap, const float rpn_positive_overlap,
const float rpn_negative_overlap, std::vector<int>* fg_inds, const float rpn_negative_overlap, std::vector<int>* fg_inds,
std::vector<int>* bg_inds, std::vector<int>* tgt_lbl, std::vector<int>* bg_inds, std::vector<int>* tgt_lbl,
std::vector<int>* fg_fake, std::vector<T>* bbox_inside_weight,
std::minstd_rand engine, bool use_random) { std::minstd_rand engine, bool use_random) {
float epsilon = 0.00001; float epsilon = 0.00001;
int anchor_num = anchor_to_gt_max.dims()[0]; int anchor_num = anchor_to_gt_max.dims()[0];
...@@ -201,12 +206,12 @@ void ScoreAssign(const T* anchor_by_gt_overlap_data, ...@@ -201,12 +206,12 @@ void ScoreAssign(const T* anchor_by_gt_overlap_data,
// Reservoir Sampling // Reservoir Sampling
int fg_num = static_cast<int>(rpn_fg_fraction * rpn_batch_size_per_im); int fg_num = static_cast<int>(rpn_fg_fraction * rpn_batch_size_per_im);
ReservoirSampling(fg_num, &fg_inds_fake, engine, use_random); ReservoirSampling(fg_num, &fg_inds_fake, engine, use_random);
fg_num = static_cast<int>(fg_inds_fake.size()); int fg_fake_num = static_cast<int>(fg_inds_fake.size());
for (int64_t i = 0; i < fg_num; ++i) { for (int64_t i = 0; i < fg_fake_num; ++i) {
target_label[fg_inds_fake[i]] = 1; target_label[fg_inds_fake[i]] = 1;
} }
int bg_num = rpn_batch_size_per_im - fg_num; int bg_num = rpn_batch_size_per_im - fg_fake_num;
for (int64_t i = 0; i < anchor_num; ++i) { for (int64_t i = 0; i < anchor_num; ++i) {
if (anchor_to_gt_max_data[i] < rpn_negative_overlap) { if (anchor_to_gt_max_data[i] < rpn_negative_overlap) {
bg_inds_fake.push_back(i); bg_inds_fake.push_back(i);
...@@ -214,12 +219,28 @@ void ScoreAssign(const T* anchor_by_gt_overlap_data, ...@@ -214,12 +219,28 @@ void ScoreAssign(const T* anchor_by_gt_overlap_data,
} }
ReservoirSampling(bg_num, &bg_inds_fake, engine, use_random); ReservoirSampling(bg_num, &bg_inds_fake, engine, use_random);
bg_num = static_cast<int>(bg_inds_fake.size()); bg_num = static_cast<int>(bg_inds_fake.size());
int fake_num = 0;
for (int64_t i = 0; i < bg_num; ++i) { for (int64_t i = 0; i < bg_num; ++i) {
// fg fake found
if (target_label[bg_inds_fake[i]] == 1) {
fake_num++;
fg_fake->emplace_back(fg_inds_fake[0]);
for (int j = 0; j < 4; ++j) {
bbox_inside_weight->emplace_back(T(0.));
}
}
target_label[bg_inds_fake[i]] = 0; target_label[bg_inds_fake[i]] = 0;
} }
for (int64_t i = 0; i < (fg_fake_num - fake_num) * 4; ++i) {
bbox_inside_weight->emplace_back(T(1.));
}
for (int64_t i = 0; i < anchor_num; ++i) { for (int64_t i = 0; i < anchor_num; ++i) {
if (target_label[i] == 1) fg_inds->emplace_back(i); if (target_label[i] == 1) {
fg_inds->emplace_back(i);
fg_fake->emplace_back(i);
}
if (target_label[i] == 0) bg_inds->emplace_back(i); if (target_label[i] == 0) bg_inds->emplace_back(i);
} }
fg_num = fg_inds->size(); fg_num = fg_inds->size();
...@@ -248,7 +269,8 @@ std::vector<Tensor> SampleRpnFgBgGt(const platform::CPUDeviceContext& ctx, ...@@ -248,7 +269,8 @@ std::vector<Tensor> SampleRpnFgBgGt(const platform::CPUDeviceContext& ctx,
std::vector<int> bg_inds; std::vector<int> bg_inds;
std::vector<int> gt_inds; std::vector<int> gt_inds;
std::vector<int> tgt_lbl; std::vector<int> tgt_lbl;
std::vector<int> fg_fake;
std::vector<T> bbox_inside_weight;
// Calculate the max IoU between anchors and gt boxes // Calculate the max IoU between anchors and gt boxes
// Map from anchor to gt box that has highest overlap // Map from anchor to gt box that has highest overlap
auto place = ctx.GetPlace(); auto place = ctx.GetPlace();
...@@ -275,32 +297,37 @@ std::vector<Tensor> SampleRpnFgBgGt(const platform::CPUDeviceContext& ctx, ...@@ -275,32 +297,37 @@ std::vector<Tensor> SampleRpnFgBgGt(const platform::CPUDeviceContext& ctx,
// Follow the Faster RCNN's implementation // Follow the Faster RCNN's implementation
ScoreAssign(anchor_by_gt_overlap_data, anchor_to_gt_max, gt_to_anchor_max, ScoreAssign(anchor_by_gt_overlap_data, anchor_to_gt_max, gt_to_anchor_max,
rpn_batch_size_per_im, rpn_fg_fraction, rpn_positive_overlap, rpn_batch_size_per_im, rpn_fg_fraction, rpn_positive_overlap,
rpn_negative_overlap, &fg_inds, &bg_inds, &tgt_lbl, engine, rpn_negative_overlap, &fg_inds, &bg_inds, &tgt_lbl, &fg_fake,
use_random); &bbox_inside_weight, engine, use_random);
int fg_num = fg_inds.size(); int fg_num = fg_inds.size();
int bg_num = bg_inds.size(); int bg_num = bg_inds.size();
gt_inds.reserve(fg_num); int fg_fake_num = fg_fake.size();
for (int i = 0; i < fg_num; ++i) { gt_inds.reserve(fg_fake_num);
gt_inds.emplace_back(argmax[fg_inds[i]]); for (int i = 0; i < fg_fake_num; ++i) {
gt_inds.emplace_back(argmax[fg_fake[i]]);
} }
Tensor loc_index_t, score_index_t, tgt_lbl_t, gt_inds_t, bbox_inside_weight_t;
Tensor loc_index_t, score_index_t, tgt_lbl_t, gt_inds_t; int* loc_index_data = loc_index_t.mutable_data<int>({fg_fake_num}, place);
int* loc_index_data = loc_index_t.mutable_data<int>({fg_num}, place);
int* score_index_data = int* score_index_data =
score_index_t.mutable_data<int>({fg_num + bg_num}, place); score_index_t.mutable_data<int>({fg_num + bg_num}, place);
int* tgt_lbl_data = tgt_lbl_t.mutable_data<int>({fg_num + bg_num}, place); int* tgt_lbl_data = tgt_lbl_t.mutable_data<int>({fg_num + bg_num}, place);
int* gt_inds_data = gt_inds_t.mutable_data<int>({fg_num}, place); int* gt_inds_data = gt_inds_t.mutable_data<int>({fg_fake_num}, place);
std::copy(fg_inds.begin(), fg_inds.end(), loc_index_data); T* bbox_inside_weight_data =
bbox_inside_weight_t.mutable_data<T>({fg_fake_num, 4}, place);
std::copy(fg_fake.begin(), fg_fake.end(), loc_index_data);
std::copy(fg_inds.begin(), fg_inds.end(), score_index_data); std::copy(fg_inds.begin(), fg_inds.end(), score_index_data);
std::copy(bg_inds.begin(), bg_inds.end(), score_index_data + fg_num); std::copy(bg_inds.begin(), bg_inds.end(), score_index_data + fg_num);
std::copy(tgt_lbl.begin(), tgt_lbl.end(), tgt_lbl_data); std::copy(tgt_lbl.begin(), tgt_lbl.end(), tgt_lbl_data);
std::copy(gt_inds.begin(), gt_inds.end(), gt_inds_data); std::copy(gt_inds.begin(), gt_inds.end(), gt_inds_data);
std::copy(bbox_inside_weight.begin(), bbox_inside_weight.end(),
bbox_inside_weight_data);
std::vector<Tensor> loc_score_tgtlbl_gt; std::vector<Tensor> loc_score_tgtlbl_gt;
loc_score_tgtlbl_gt.emplace_back(loc_index_t); loc_score_tgtlbl_gt.emplace_back(loc_index_t);
loc_score_tgtlbl_gt.emplace_back(score_index_t); loc_score_tgtlbl_gt.emplace_back(score_index_t);
loc_score_tgtlbl_gt.emplace_back(tgt_lbl_t); loc_score_tgtlbl_gt.emplace_back(tgt_lbl_t);
loc_score_tgtlbl_gt.emplace_back(gt_inds_t); loc_score_tgtlbl_gt.emplace_back(gt_inds_t);
loc_score_tgtlbl_gt.emplace_back(bbox_inside_weight_t);
return loc_score_tgtlbl_gt; return loc_score_tgtlbl_gt;
} }
...@@ -318,6 +345,7 @@ class RpnTargetAssignKernel : public framework::OpKernel<T> { ...@@ -318,6 +345,7 @@ class RpnTargetAssignKernel : public framework::OpKernel<T> {
auto* score_index = context.Output<LoDTensor>("ScoreIndex"); auto* score_index = context.Output<LoDTensor>("ScoreIndex");
auto* tgt_bbox = context.Output<LoDTensor>("TargetBBox"); auto* tgt_bbox = context.Output<LoDTensor>("TargetBBox");
auto* tgt_lbl = context.Output<LoDTensor>("TargetLabel"); auto* tgt_lbl = context.Output<LoDTensor>("TargetLabel");
auto* bbox_inside_weight = context.Output<LoDTensor>("BBox_inside_weight");
PADDLE_ENFORCE_EQ(gt_boxes->lod().size(), 1UL, PADDLE_ENFORCE_EQ(gt_boxes->lod().size(), 1UL,
"RpnTargetAssignOp gt_boxes needs 1 level of LoD"); "RpnTargetAssignOp gt_boxes needs 1 level of LoD");
...@@ -340,7 +368,7 @@ class RpnTargetAssignKernel : public framework::OpKernel<T> { ...@@ -340,7 +368,7 @@ class RpnTargetAssignKernel : public framework::OpKernel<T> {
score_index->mutable_data<int>({max_num}, place); score_index->mutable_data<int>({max_num}, place);
tgt_bbox->mutable_data<T>({max_num, 4}, place); tgt_bbox->mutable_data<T>({max_num, 4}, place);
tgt_lbl->mutable_data<int>({max_num, 1}, place); tgt_lbl->mutable_data<int>({max_num, 1}, place);
bbox_inside_weight->mutable_data<T>({max_num, 4}, place);
auto& dev_ctx = context.device_context<platform::CPUDeviceContext>(); auto& dev_ctx = context.device_context<platform::CPUDeviceContext>();
std::random_device rnd; std::random_device rnd;
...@@ -394,6 +422,7 @@ class RpnTargetAssignKernel : public framework::OpKernel<T> { ...@@ -394,6 +422,7 @@ class RpnTargetAssignKernel : public framework::OpKernel<T> {
Tensor sampled_score_index = loc_score_tgtlbl_gt[1]; Tensor sampled_score_index = loc_score_tgtlbl_gt[1];
Tensor sampled_tgtlbl = loc_score_tgtlbl_gt[2]; Tensor sampled_tgtlbl = loc_score_tgtlbl_gt[2];
Tensor sampled_gt_index = loc_score_tgtlbl_gt[3]; Tensor sampled_gt_index = loc_score_tgtlbl_gt[3];
Tensor sampled_bbox_inside_weight = loc_score_tgtlbl_gt[4];
int loc_num = sampled_loc_index.dims()[0]; int loc_num = sampled_loc_index.dims()[0];
int score_num = sampled_score_index.dims()[0]; int score_num = sampled_score_index.dims()[0];
...@@ -432,6 +461,8 @@ class RpnTargetAssignKernel : public framework::OpKernel<T> { ...@@ -432,6 +461,8 @@ class RpnTargetAssignKernel : public framework::OpKernel<T> {
AppendRpns<int>(score_index, total_score_num, &sampled_score_index_unmap); AppendRpns<int>(score_index, total_score_num, &sampled_score_index_unmap);
AppendRpns<T>(tgt_bbox, total_loc_num * 4, &sampled_tgt_bbox); AppendRpns<T>(tgt_bbox, total_loc_num * 4, &sampled_tgt_bbox);
AppendRpns<int>(tgt_lbl, total_score_num, &sampled_tgtlbl); AppendRpns<int>(tgt_lbl, total_score_num, &sampled_tgtlbl);
AppendRpns<T>(bbox_inside_weight, total_loc_num * 4,
&sampled_bbox_inside_weight);
total_loc_num += loc_num; total_loc_num += loc_num;
total_score_num += score_num; total_score_num += score_num;
...@@ -448,10 +479,12 @@ class RpnTargetAssignKernel : public framework::OpKernel<T> { ...@@ -448,10 +479,12 @@ class RpnTargetAssignKernel : public framework::OpKernel<T> {
score_index->set_lod(loc_score); score_index->set_lod(loc_score);
tgt_bbox->set_lod(lod_loc); tgt_bbox->set_lod(lod_loc);
tgt_lbl->set_lod(loc_score); tgt_lbl->set_lod(loc_score);
bbox_inside_weight->set_lod(lod_loc);
loc_index->Resize({total_loc_num}); loc_index->Resize({total_loc_num});
score_index->Resize({total_score_num}); score_index->Resize({total_score_num});
tgt_bbox->Resize({total_loc_num, 4}); tgt_bbox->Resize({total_loc_num, 4});
tgt_lbl->Resize({total_score_num, 1}); tgt_lbl->Resize({total_score_num, 1});
bbox_inside_weight->Resize({total_loc_num, 4});
} }
}; };
...@@ -514,6 +547,9 @@ class RpnTargetAssignOpMaker : public framework::OpProtoAndCheckerMaker { ...@@ -514,6 +547,9 @@ class RpnTargetAssignOpMaker : public framework::OpProtoAndCheckerMaker {
"TargetLabel", "TargetLabel",
"(Tensor<int>), The target labels of each anchor with shape " "(Tensor<int>), The target labels of each anchor with shape "
"[F + B, 1], F and B are sampled foreground and backgroud number."); "[F + B, 1], F and B are sampled foreground and backgroud number.");
AddOutput("BBox_inside_weight",
"(Tensor), The bbox inside weight with shape "
"[F, 4], F is the sampled foreground number.");
AddComment(R"DOC( AddComment(R"DOC(
This operator can be, for a given set of ground truth bboxes and the This operator can be, for a given set of ground truth bboxes and the
anchors, to assign classification and regression targets to each prediction. anchors, to assign classification and regression targets to each prediction.
......
...@@ -116,8 +116,8 @@ def rpn_target_assign(bbox_pred, ...@@ -116,8 +116,8 @@ def rpn_target_assign(bbox_pred,
Returns: Returns:
tuple: tuple:
A tuple(predicted_scores, predicted_location, target_label, A tuple(predicted_scores, predicted_location, target_label,
target_bbox) is returned. The predicted_scores and target_bbox, bbox_inside_weight) is returned. The predicted_scores
predicted_location is the predicted result of the RPN. and predicted_location is the predicted result of the RPN.
The target_label and target_bbox is the ground truth, The target_label and target_bbox is the ground truth,
respectively. The predicted_location is a 2D Tensor with shape respectively. The predicted_location is a 2D Tensor with shape
[F, 4], and the shape of target_bbox is same as the shape of [F, 4], and the shape of target_bbox is same as the shape of
...@@ -126,6 +126,8 @@ def rpn_target_assign(bbox_pred, ...@@ -126,6 +126,8 @@ def rpn_target_assign(bbox_pred,
[F + B, 1], and the shape of target_label is same as the shape [F + B, 1], and the shape of target_label is same as the shape
of the predicted_scores, B is the number of the background of the predicted_scores, B is the number of the background
anchors, the F and B is depends on the input of this operator. anchors, the F and B is depends on the input of this operator.
Bbox_inside_weight represents whether the predicted loc is fake_fg
or not and the shape is [F, 4].
Examples: Examples:
.. code-block:: python .. code-block:: python
...@@ -138,7 +140,7 @@ def rpn_target_assign(bbox_pred, ...@@ -138,7 +140,7 @@ def rpn_target_assign(bbox_pred,
append_batch_size=False, dtype='float32') append_batch_size=False, dtype='float32')
gt_boxes = layers.data(name='gt_boxes', shape=[10, 4], gt_boxes = layers.data(name='gt_boxes', shape=[10, 4],
append_batch_size=False, dtype='float32') append_batch_size=False, dtype='float32')
loc_pred, score_pred, loc_target, score_target = loc_pred, score_pred, loc_target, score_target, bbox_inside_weight =
fluid.layers.rpn_target_assign(bbox_pred=bbox_pred, fluid.layers.rpn_target_assign(bbox_pred=bbox_pred,
cls_logits=cls_logits, cls_logits=cls_logits,
anchor_box=anchor_box, anchor_box=anchor_box,
...@@ -151,6 +153,7 @@ def rpn_target_assign(bbox_pred, ...@@ -151,6 +153,7 @@ def rpn_target_assign(bbox_pred,
score_index = helper.create_tmp_variable(dtype='int32') score_index = helper.create_tmp_variable(dtype='int32')
target_label = helper.create_tmp_variable(dtype='int32') target_label = helper.create_tmp_variable(dtype='int32')
target_bbox = helper.create_tmp_variable(dtype=anchor_box.dtype) target_bbox = helper.create_tmp_variable(dtype=anchor_box.dtype)
bbox_inside_weight = helper.create_tmp_variable(dtype=anchor_box.dtype)
helper.append_op( helper.append_op(
type="rpn_target_assign", type="rpn_target_assign",
inputs={ inputs={
...@@ -163,7 +166,8 @@ def rpn_target_assign(bbox_pred, ...@@ -163,7 +166,8 @@ def rpn_target_assign(bbox_pred,
'LocationIndex': loc_index, 'LocationIndex': loc_index,
'ScoreIndex': score_index, 'ScoreIndex': score_index,
'TargetLabel': target_label, 'TargetLabel': target_label,
'TargetBBox': target_bbox 'TargetBBox': target_bbox,
'BBox_inside_weight': bbox_inside_weight
}, },
attrs={ attrs={
'rpn_batch_size_per_im': rpn_batch_size_per_im, 'rpn_batch_size_per_im': rpn_batch_size_per_im,
...@@ -178,13 +182,14 @@ def rpn_target_assign(bbox_pred, ...@@ -178,13 +182,14 @@ def rpn_target_assign(bbox_pred,
score_index.stop_gradient = True score_index.stop_gradient = True
target_label.stop_gradient = True target_label.stop_gradient = True
target_bbox.stop_gradient = True target_bbox.stop_gradient = True
bbox_inside_weight.stop_gradient = True
cls_logits = nn.reshape(x=cls_logits, shape=(-1, 1)) cls_logits = nn.reshape(x=cls_logits, shape=(-1, 1))
bbox_pred = nn.reshape(x=bbox_pred, shape=(-1, 4)) bbox_pred = nn.reshape(x=bbox_pred, shape=(-1, 4))
predicted_cls_logits = nn.gather(cls_logits, score_index) predicted_cls_logits = nn.gather(cls_logits, score_index)
predicted_bbox_pred = nn.gather(bbox_pred, loc_index) predicted_bbox_pred = nn.gather(bbox_pred, loc_index)
return predicted_cls_logits, predicted_bbox_pred, target_label, target_bbox return predicted_cls_logits, predicted_bbox_pred, target_label, target_bbox, bbox_inside_weight
def detection_output(loc, def detection_output(loc,
......
...@@ -301,7 +301,7 @@ class TestRpnTargetAssign(unittest.TestCase): ...@@ -301,7 +301,7 @@ class TestRpnTargetAssign(unittest.TestCase):
dtype='float32', dtype='float32',
lod_level=1, lod_level=1,
append_batch_size=False) append_batch_size=False)
pred_scores, pred_loc, tgt_lbl, tgt_bbox = layers.rpn_target_assign( pred_scores, pred_loc, tgt_lbl, tgt_bbox, bbox_inside_weight = layers.rpn_target_assign(
bbox_pred=bbox_pred, bbox_pred=bbox_pred,
cls_logits=cls_logits, cls_logits=cls_logits,
anchor_box=anchor_box, anchor_box=anchor_box,
...@@ -313,12 +313,14 @@ class TestRpnTargetAssign(unittest.TestCase): ...@@ -313,12 +313,14 @@ class TestRpnTargetAssign(unittest.TestCase):
rpn_straddle_thresh=0.0, rpn_straddle_thresh=0.0,
rpn_fg_fraction=0.5, rpn_fg_fraction=0.5,
rpn_positive_overlap=0.7, rpn_positive_overlap=0.7,
rpn_negative_overlap=0.3) rpn_negative_overlap=0.3,
use_random=False)
self.assertIsNotNone(pred_scores) self.assertIsNotNone(pred_scores)
self.assertIsNotNone(pred_loc) self.assertIsNotNone(pred_loc)
self.assertIsNotNone(tgt_lbl) self.assertIsNotNone(tgt_lbl)
self.assertIsNotNone(tgt_bbox) self.assertIsNotNone(tgt_bbox)
self.assertIsNotNone(bbox_inside_weight)
assert pred_scores.shape[1] == 1 assert pred_scores.shape[1] == 1
assert pred_loc.shape[1] == 4 assert pred_loc.shape[1] == 4
assert pred_loc.shape[1] == tgt_bbox.shape[1] assert pred_loc.shape[1] == tgt_bbox.shape[1]
......
...@@ -50,8 +50,10 @@ def rpn_target_assign(anchor_by_gt_overlap, ...@@ -50,8 +50,10 @@ def rpn_target_assign(anchor_by_gt_overlap,
fg_inds, size=(len(fg_inds) - num_fg), replace=False) fg_inds, size=(len(fg_inds) - num_fg), replace=False)
else: else:
disable_inds = fg_inds[num_fg:] disable_inds = fg_inds[num_fg:]
labels[disable_inds] = -1 labels[disable_inds] = -1
fg_inds = np.where(labels == 1)[0] fg_inds = np.where(labels == 1)[0]
bbox_inside_weight = np.zeros((len(fg_inds), 4), dtype=np.float32)
num_bg = rpn_batch_size_per_im - np.sum(labels == 1) num_bg = rpn_batch_size_per_im - np.sum(labels == 1)
bg_inds = np.where(anchor_to_gt_max < rpn_negative_overlap)[0] bg_inds = np.where(anchor_to_gt_max < rpn_negative_overlap)[0]
...@@ -59,18 +61,27 @@ def rpn_target_assign(anchor_by_gt_overlap, ...@@ -59,18 +61,27 @@ def rpn_target_assign(anchor_by_gt_overlap,
enable_inds = bg_inds[np.random.randint(len(bg_inds), size=num_bg)] enable_inds = bg_inds[np.random.randint(len(bg_inds), size=num_bg)]
else: else:
enable_inds = bg_inds[:num_bg] enable_inds = bg_inds[:num_bg]
fg_fake_inds = np.array([], np.int32)
fg_value = np.array([fg_inds[0]], np.int32)
fake_num = 0
for bg_id in enable_inds:
if bg_id in fg_inds:
fake_num += 1
fg_fake_inds = np.hstack([fg_fake_inds, fg_value])
labels[enable_inds] = 0 labels[enable_inds] = 0
bbox_inside_weight[fake_num:, :] = 1
fg_inds = np.where(labels == 1)[0] fg_inds = np.where(labels == 1)[0]
bg_inds = np.where(labels == 0)[0] bg_inds = np.where(labels == 0)[0]
loc_index = np.hstack([fg_fake_inds, fg_inds])
loc_index = fg_inds score_index = np.hstack([fg_inds, bg_inds])
score_index = np.hstack((fg_inds, bg_inds))
labels = labels[score_index] labels = labels[score_index]
assert not np.any(labels == -1), "Wrong labels with -1" assert not np.any(labels == -1), "Wrong labels with -1"
gt_inds = anchor_to_gt_argmax[fg_inds] gt_inds = anchor_to_gt_argmax[loc_index]
return loc_index, score_index, labels, gt_inds return loc_index, score_index, labels, gt_inds, bbox_inside_weight
def get_anchor(n, c, h, w): def get_anchor(n, c, h, w):
...@@ -123,9 +134,12 @@ def rpn_target_assign_in_python(all_anchors, ...@@ -123,9 +134,12 @@ def rpn_target_assign_in_python(all_anchors,
gt_boxes_slice = gt_boxes_slice[not_crowd_inds] gt_boxes_slice = gt_boxes_slice[not_crowd_inds]
iou = _bbox_overlaps(inside_anchors, gt_boxes_slice) iou = _bbox_overlaps(inside_anchors, gt_boxes_slice)
loc_inds, score_inds, labels, gt_inds = rpn_target_assign( loc_inds, score_inds, labels, gt_inds, bbox_inside_weight = \
iou, rpn_batch_size_per_im, rpn_positive_overlap, rpn_target_assign(iou, rpn_batch_size_per_im,
rpn_negative_overlap, rpn_fg_fraction, use_random) rpn_positive_overlap,
rpn_negative_overlap,
rpn_fg_fraction,
use_random)
# unmap to all anchor # unmap to all anchor
loc_inds = inds_inside[loc_inds] loc_inds = inds_inside[loc_inds]
score_inds = inds_inside[score_inds] score_inds = inds_inside[score_inds]
...@@ -139,6 +153,7 @@ def rpn_target_assign_in_python(all_anchors, ...@@ -139,6 +153,7 @@ def rpn_target_assign_in_python(all_anchors,
score_indexes = score_inds score_indexes = score_inds
tgt_labels = labels tgt_labels = labels
tgt_bboxes = box_deltas tgt_bboxes = box_deltas
bbox_inside_weights = bbox_inside_weight
else: else:
loc_indexes = np.concatenate( loc_indexes = np.concatenate(
[loc_indexes, loc_inds + i * anchor_num]) [loc_indexes, loc_inds + i * anchor_num])
...@@ -146,8 +161,10 @@ def rpn_target_assign_in_python(all_anchors, ...@@ -146,8 +161,10 @@ def rpn_target_assign_in_python(all_anchors,
[score_indexes, score_inds + i * anchor_num]) [score_indexes, score_inds + i * anchor_num])
tgt_labels = np.concatenate([tgt_labels, labels]) tgt_labels = np.concatenate([tgt_labels, labels])
tgt_bboxes = np.vstack([tgt_bboxes, box_deltas]) tgt_bboxes = np.vstack([tgt_bboxes, box_deltas])
bbox_inside_weights = np.vstack([bbox_inside_weights, \
bbox_inside_weight])
return loc_indexes, score_indexes, tgt_bboxes, tgt_labels return loc_indexes, score_indexes, tgt_bboxes, tgt_labels, bbox_inside_weights
class TestRpnTargetAssignOp(OpTest): class TestRpnTargetAssignOp(OpTest):
...@@ -182,9 +199,11 @@ class TestRpnTargetAssignOp(OpTest): ...@@ -182,9 +199,11 @@ class TestRpnTargetAssignOp(OpTest):
rpn_fg_fraction = 0.5 rpn_fg_fraction = 0.5
use_random = False use_random = False
loc_index, score_index, tgt_bbox, labels = rpn_target_assign_in_python( loc_index, score_index, tgt_bbox, labels, bbox_inside_weights = \
all_anchors, gt_boxes, is_crowd, im_info, lod, rpn_straddle_thresh, rpn_target_assign_in_python(all_anchors, gt_boxes, is_crowd,
rpn_batch_size_per_im, rpn_positive_overlap, rpn_negative_overlap, im_info, lod, rpn_straddle_thresh,
rpn_batch_size_per_im, rpn_positive_overlap,
rpn_negative_overlap,
rpn_fg_fraction, use_random) rpn_fg_fraction, use_random)
labels = labels[:, np.newaxis] labels = labels[:, np.newaxis]
...@@ -207,7 +226,8 @@ class TestRpnTargetAssignOp(OpTest): ...@@ -207,7 +226,8 @@ class TestRpnTargetAssignOp(OpTest):
'LocationIndex': loc_index.astype('int32'), 'LocationIndex': loc_index.astype('int32'),
'ScoreIndex': score_index.astype('int32'), 'ScoreIndex': score_index.astype('int32'),
'TargetBBox': tgt_bbox.astype('float32'), 'TargetBBox': tgt_bbox.astype('float32'),
'TargetLabel': labels.astype('int32') 'TargetLabel': labels.astype('int32'),
'BBox_inside_weight': bbox_inside_weights.astype('float32')
} }
def test_check_output(self): def test_check_output(self):
......
Markdown is supported
0% .
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