From 5c20b55eadeebb9cde002a4117723cbf22b4f927 Mon Sep 17 00:00:00 2001 From: wqz960 <362379625@qq.com> Date: Mon, 20 Jul 2020 03:28:10 +0000 Subject: [PATCH] modify position --- .../feature_visualization_input.jpg | Bin .../feature_visualization_output.jpg | Bin .../feature_visiualization}/get_started.md | 9 +- .../download_resnet50_pretrained.sh | 2 + tools/feature_maps_visualization/fm_vis.py | 9 +- tools/feature_maps_visualization/resnet.py | 215 ++++++++++++++++++ 6 files changed, 223 insertions(+), 12 deletions(-) rename tools/feature_maps_visualization/test.jpg => docs/images/feature_maps/feature_visualization_input.jpg (100%) rename tools/feature_maps_visualization/fm.jpg => docs/images/feature_maps/feature_visualization_output.jpg (100%) rename {tools/feature_maps_visualization => docs/zh_CN/feature_visiualization}/get_started.md (78%) create mode 100644 tools/feature_maps_visualization/download_resnet50_pretrained.sh create mode 100644 tools/feature_maps_visualization/resnet.py diff --git a/tools/feature_maps_visualization/test.jpg b/docs/images/feature_maps/feature_visualization_input.jpg similarity index 100% rename from tools/feature_maps_visualization/test.jpg rename to docs/images/feature_maps/feature_visualization_input.jpg diff --git a/tools/feature_maps_visualization/fm.jpg b/docs/images/feature_maps/feature_visualization_output.jpg similarity index 100% rename from tools/feature_maps_visualization/fm.jpg rename to docs/images/feature_maps/feature_visualization_output.jpg diff --git a/tools/feature_maps_visualization/get_started.md b/docs/zh_CN/feature_visiualization/get_started.md similarity index 78% rename from tools/feature_maps_visualization/get_started.md rename to docs/zh_CN/feature_visiualization/get_started.md index e11fe968..a287aad8 100644 --- a/tools/feature_maps_visualization/get_started.md +++ b/docs/zh_CN/feature_visiualization/get_started.md @@ -6,7 +6,7 @@ ## 二、准备工作 -首先我们需要选定研究的模型，本文设定ResNet50作为研究模型，将resnet.py从[模型库](../../ppcls/modeling/architecture/)拷贝到当前目录下，并下载预训练模型[预训练模型](../../docs/zh_CN/models/models_intro), 复制resnet50的模型链接，使用下列命令下载并解压预训练模型。 +首先需要选定研究的模型，本文设定ResNet50作为研究模型，将resnet.py从[模型库](../../../ppcls/modeling/architecture/)拷贝到当前目录下，并下载预训练模型[预训练模型](../../zh_CN/models/models_intro), 复制resnet50的模型链接，使用下列命令下载并解压预训练模型。 ```bash wget The Link for Pretrained Model @@ -53,18 +53,17 @@ python tools/feature_maps_visualization/fm_vis.py -i the image you want to test ``` 参数说明： + `-i`：待预测的图片文件路径，如 `./test.jpeg` -+ `-c`：特征图维度，如 `./resnet50-vd/model` ++ `-c`：特征图维度，如 `./resnet50_vd/model` + `-p`：权重文件路径，如 `./ResNet50_pretrained/` + `--show`：是否展示图片，默认值 False -+ `--save`：是否保存图片，默认值：True + `--save_path`：保存路径，如：`./tools/` + `--use_gpu`：是否使用 GPU 预测，默认值：True ## 四、结果 输入图片： -![](../../tools/feature_maps_visualization/test.jpg) +![](../../../tools/feature_maps_visualization/test.jpg) 输出特征图： -![](../../tools/feature_maps_visualization/fm.jpg) +![](../../../tools/feature_maps_visualization/fm.jpg) diff --git a/tools/feature_maps_visualization/download_resnet50_pretrained.sh b/tools/feature_maps_visualization/download_resnet50_pretrained.sh new file mode 100644 index 00000000..286c2400 --- /dev/null +++ b/tools/feature_maps_visualization/download_resnet50_pretrained.sh @@ -0,0 +1,2 @@ +wget https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_pretrained.tar +tar -xf ResNet50_pretrained.tar \ No newline at end of file diff --git a/tools/feature_maps_visualization/fm_vis.py b/tools/feature_maps_visualization/fm_vis.py index efb983ca..1731313b 100644 --- a/tools/feature_maps_visualization/fm_vis.py +++ b/tools/feature_maps_visualization/fm_vis.py @@ -28,7 +28,6 @@ def parse_args(): parser.add_argument("-c", "--channel_num", type=int) parser.add_argument("-p", "--pretrained_model", type=str) parser.add_argument("--show", type=str2bool, default=False) - parser.add_argument("--save", type=str2bool, default=True) parser.add_argument("--save_path", type=str) parser.add_argument("--use_gpu", type=str2bool, default=True) @@ -66,9 +65,7 @@ def main(): place = fluid.CUDAPlace(gpu_id) else: place = fluid.CPUPlace() - fm = None - - print(args.pretrained_model) + pre_weights_dict = fluid.load_program_state(args.pretrained_model) with fluid.dygraph.guard(place): net = ResNet50() @@ -83,12 +80,10 @@ def main(): net.set_dict(pre_weights_dict_new) net.eval() _, fm = net(data) - assert args.channel_num >= 0 and args.channel_num <= fm.shape[1], "the channel is out of the range" + assert args.channel_num >= 0 and args.channel_num <= fm.shape[1], "the channel is out of the range, should be in {} but got {}".format([0, fm.shape[1]], args.channel_num) fm = (np.squeeze(fm[0][args.channel_num].numpy())*255).astype(np.uint8) - print(fm) if fm is not None: if args.save: - print(args.save_path) cv2.imwrite(args.save_path, fm) if args.show: cv2.show(fm) diff --git a/tools/feature_maps_visualization/resnet.py b/tools/feature_maps_visualization/resnet.py new file mode 100644 index 00000000..d3f230da --- /dev/null +++ b/tools/feature_maps_visualization/resnet.py @@ -0,0 +1,215 @@ +import numpy as np +import argparse +import ast +import paddle +import paddle.fluid as fluid +from paddle.fluid.param_attr import ParamAttr +from paddle.fluid.layer_helper import LayerHelper +from paddle.fluid.dygraph.nn import Conv2D, Pool2D, BatchNorm, Linear +from paddle.fluid.dygraph.base import to_variable + +from paddle.fluid import framework + +import math +import sys +import time + +class ConvBNLayer(fluid.dygraph.Layer): + def __init__(self, + num_channels, + num_filters, + filter_size, + stride=1, + groups=1, + act=None, + name=None): + super(ConvBNLayer, self).__init__() + + self._conv = Conv2D( + num_channels=num_channels, + num_filters=num_filters, + filter_size=filter_size, + stride=stride, + padding=(filter_size - 1) // 2, + groups=groups, + act=None, + param_attr=ParamAttr(name=name + "_weights"), + bias_attr=False) + if name == "conv1": + bn_name = "bn_" + name + else: + bn_name = "bn" + name[3:] + self._batch_norm = BatchNorm(num_filters, + act=act, + param_attr=ParamAttr(name=bn_name + '_scale'), + bias_attr=ParamAttr(bn_name + '_offset'), + moving_mean_name=bn_name + '_mean', + moving_variance_name=bn_name + '_variance') + + def forward(self, inputs): + y = self._conv(inputs) + y = self._batch_norm(y) + return y + + +class BottleneckBlock(fluid.dygraph.Layer): + def __init__(self, + num_channels, + num_filters, + stride, + shortcut=True, + name=None): + super(BottleneckBlock, self).__init__() + + self.conv0 = ConvBNLayer( + num_channels=num_channels, + num_filters=num_filters, + filter_size=1, + act='relu', + name=name+"_branch2a") + self.conv1 = ConvBNLayer( + num_channels=num_filters, + num_filters=num_filters, + filter_size=3, + stride=stride, + act='relu', + name=name+"_branch2b") + self.conv2 = ConvBNLayer( + num_channels=num_filters, + num_filters=num_filters * 4, + filter_size=1, + act=None, + name=name+"_branch2c") + + if not shortcut: + self.short = ConvBNLayer( + num_channels=num_channels, + num_filters=num_filters * 4, + filter_size=1, + stride=stride, + name=name + "_branch1") + + self.shortcut = shortcut + + self._num_channels_out = num_filters * 4 + + def forward(self, inputs): + y = self.conv0(inputs) + conv1 = self.conv1(y) + conv2 = self.conv2(conv1) + + if self.shortcut: + short = inputs + else: + short = self.short(inputs) + + y = fluid.layers.elementwise_add(x=short, y=conv2) + + layer_helper = LayerHelper(self.full_name(), act='relu') + return layer_helper.append_activation(y) + + +class ResNet(fluid.dygraph.Layer): + def __init__(self, layers=50, class_dim=1000): + super(ResNet, self).__init__() + + self.layers = layers + supported_layers = [50, 101, 152] + assert layers in supported_layers, \ + "supported layers are {} but input layer is {}".format(supported_layers, layers) + self.fm = None + + if layers == 50: + depth = [3, 4, 6, 3] + elif layers == 101: + depth = [3, 4, 23, 3] + elif layers == 152: + depth = [3, 8, 36, 3] + num_channels = [64, 256, 512, 1024] + num_filters = [64, 128, 256, 512] + + self.conv = ConvBNLayer( + num_channels=3, + num_filters=64, + filter_size=7, + stride=2, + act='relu', + name="conv1") + self.pool2d_max = Pool2D( + pool_size=3, + pool_stride=2, + pool_padding=1, + pool_type='max') + + self.bottleneck_block_list = [] + for block in range(len(depth)): + shortcut = False + for i in range(depth[block]): + if layers in [101, 152] and block == 2: + if i == 0: + conv_name="res"+str(block+2)+"a" + else: + conv_name="res"+str(block+2)+"b"+str(i) + else: + conv_name="res"+str(block+2)+chr(97+i) + bottleneck_block = self.add_sublayer( + 'bb_%d_%d' % (block, i), + BottleneckBlock( + num_channels=num_channels[block] + if i == 0 else num_filters[block] * 4, + num_filters=num_filters[block], + stride=2 if i == 0 and block != 0 else 1, + shortcut=shortcut, + name=conv_name)) + self.bottleneck_block_list.append(bottleneck_block) + shortcut = True + + self.pool2d_avg = Pool2D( + pool_size=7, pool_type='avg', global_pooling=True) + + self.pool2d_avg_output = num_filters[len(num_filters) - 1] * 4 * 1 * 1 + + stdv = 1.0 / math.sqrt(2048 * 1.0) + + self.out = Linear(self.pool2d_avg_output, + class_dim, + param_attr=ParamAttr( + initializer=fluid.initializer.Uniform(-stdv, stdv), name="fc_0.w_0"), + bias_attr=ParamAttr(name="fc_0.b_0")) + + def forward(self, inputs): + y = self.conv(inputs) + y = self.pool2d_max(y) + self.fm = y + for bottleneck_block in self.bottleneck_block_list: + y = bottleneck_block(y) + y = self.pool2d_avg(y) + y = fluid.layers.reshape(y, shape=[-1, self.pool2d_avg_output]) + y = self.out(y) + return y, self.fm + + +def ResNet50(**args): + model = ResNet(layers=50, **args) + return model + + +def ResNet101(**args): + model = ResNet(layers=101, **args) + return model + + +def ResNet152(**args): + model = ResNet(layers=152, **args) + return model + + +if __name__ == "__main__": + import numpy as np + place = fluid.CPUPlace() + with fluid.dygraph.guard(place): + model = ResNet50() + img = np.random.uniform(0, 255, [1, 3, 224, 224]).astype('float32') + img = fluid.dygraph.to_variable(img) + res = model(img) + print(res.shape) -- GitLab