update slim model (#2427)

* update slim model

* fix skip quant

configs/slim/README.md

@@ -87,13 +87,13 @@ python tools/export_model.py -c configs/{MODEL.yml} --slim_config configs/slim/{
 | 模型               | 压缩策略     | 输入尺寸 |   Box AP    |                             下载                             |                         模型配置文件                         |                       压缩算法配置文件                       |
 | ------------------ | ------------ | -------- | :---------: | :----------------------------------------------------------: | :----------------------------------------------------------: | :----------------------------------------------------------: |
 | YOLOv3-MobileNetV1 | baseline     | 608      |    28.8     | [下载链接](https://paddledet.bj.bcebos.com/models/yolov3_mobilenet_v1_270e_coco.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/yolov3/yolov3_mobilenet_v1_270e_coco.yml) |                            -                               |
-| YOLOv3-MobileNetV1 | 普通在线量化 | 608      | 30.3 (+1.5) | [下载链接](https://paddledet.bj.bcebos.com/models/slim/yolov3_mobilenet_v1_coco_qat.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/yolov3/yolov3_mobilenet_v1_270e_coco.yml) | [slim配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/slim/quant/yolov3_mobilenet_v1_qat.yml) |
+| YOLOv3-MobileNetV1 | 普通在线量化 | 608      | 30.5 (+1.7) | [下载链接](https://paddledet.bj.bcebos.com/models/slim/yolov3_mobilenet_v1_coco_qat.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/yolov3/yolov3_mobilenet_v1_270e_coco.yml) | [slim配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/slim/quant/yolov3_mobilenet_v1_qat.yml) |
 | YOLOv3-MobileNetV3 | baseline     | 608      |    31.4     | [下载链接](https://paddledet.bj.bcebos.com/models/yolov3_mobilenet_v3_large_270e_coco.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/yolov3/yolov3_mobilenet_v3_large_270e_coco.yml) |                              -                               |
-| YOLOv3-MobileNetV3 | PACT在线量化 | 608      | 29.5 (-1.9) | [下载链接](https://paddledet.bj.bcebos.com/models/slim/yolov3_mobilenet_v3_coco_qat.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/yolov3/yolov3_mobilenet_v3_large_270e_coco.yml) | [slim配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/slim/quant/yolov3_mobilenet_v3_qat.yml) |
+| YOLOv3-MobileNetV3 | PACT在线量化 | 608      | 29.1 (-2.3) | [下载链接](https://paddledet.bj.bcebos.com/models/slim/yolov3_mobilenet_v3_coco_qat.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/yolov3/yolov3_mobilenet_v3_large_270e_coco.yml) | [slim配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/slim/quant/yolov3_mobilenet_v3_qat.yml) |
 | YOLOv3-DarkNet53 | baseline     | 608      |    39.0     | [下载链接](https://paddledet.bj.bcebos.com/models/yolov3_darknet53_270e_coco.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/yolov3/yolov3_darknet53_270e_coco.yml) |                           -                               |
 | YOLOv3-DarkNet53 | 普通在线量化 | 608      | 38.7 (-0.3) | [下载链接](https://paddledet.bj.bcebos.com/models/slim/yolov3_darknet_coco_qat.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/yolov3/yolov3_darknet53_270e_coco.yml) | [slim配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/slim/quant/yolov3_darknet_qat.yml) |
 | SSD-MobileNet_v1    |  baseline   |   300   |  73.8  | [下载链接](https://paddledet.bj.bcebos.com/models/ssd_mobilenet_v1_300_120e_voc.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/ssd/ssd_mobilenet_v1_300_120e_voc.yml) |     -    |
-| SSD-MobileNet_v1    |  普通在线量化   |   300   |  73.1(-0.7)  | [下载链接](https://paddledet.bj.bcebos.com/models/slim/ssd_mobilenet_v1_300_voc_qat.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/ssd/ssd_mobilenet_v1_300_120e_voc.yml) | [slim配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/slim/quant/ssd_mobilenet_v1_qat.yml) |
+| SSD-MobileNet_v1    |  普通在线量化   |   300   |  72.9(-0.9)  | [下载链接](https://paddledet.bj.bcebos.com/models/slim/ssd_mobilenet_v1_300_voc_qat.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/ssd/ssd_mobilenet_v1_300_120e_voc.yml) | [slim配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/slim/quant/ssd_mobilenet_v1_qat.yml) |
 | Mask-ResNet50-FPN     |    baseline      |    (800, 1333)   |  39.2/35.6    | [下载链接](https://paddledet.bj.bcebos.com/models/mask_rcnn_r50_fpn_1x_coco.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/mask_rcnn/mask_rcnn_r50_fpn_1x_coco.yml) |  -  |
 | Mask-ResNet50-FPN     |    普通在线量化      |    (800, 1333)   |  39.7(+0.5)/35.9(+0.3)    | [下载链接](https://paddledet.bj.bcebos.com/models/slim/mask_rcnn_r50_fpn_1x_qat.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/mask_rcnn/mask_rcnn_r50_fpn_1x_coco.yml) |  [slim配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/slim/quant/mask_rcnn_r50_fpn_1x_qat.yml)  |
 

configs/slim/quant/mask_rcnn_r50_fpn_1x_qat.yml

+pretrain_weights: https://paddledet.bj.bcebos.com/models/mask_rcnn_r50_fpn_1x_coco.pdparams
+slim: QAT
+
+QAT:
+  quant_config: {
+    'weight_quantize_type': 'channel_wise_abs_max', 'activation_quantize_type': 'moving_average_abs_max',
+    'weight_bits': 8, 'activation_bits': 8, 'dtype': 'int8', 'window_size': 10000, 'moving_rate': 0.9,
+    'quantizable_layer_type': ['Conv2D', 'Linear']}
+  print_model: True
+
+
+epoch: 5
+
+LearningRate:
+  base_lr: 0.001
+  schedulers:
+  - !PiecewiseDecay
+    gamma: 0.1
+    milestones: [3, 4]
+  - !LinearWarmup
+    start_factor: 0.001
+    steps: 100

configs/slim/quant/ssd_mobilenet_v1_qat.yml

+pretrain_weights: https://paddlemodels.bj.bcebos.com/object_detection/dygraph/ssd_mobilenet_v1_300_120e_voc.pdparams
+slim: QAT
+
+QAT:
+  quant_config: {
+    'weight_quantize_type': 'channel_wise_abs_max', 'activation_quantize_type': 'moving_average_abs_max',
+    'weight_bits': 8, 'activation_bits': 8, 'dtype': 'int8', 'window_size': 10000, 'moving_rate': 0.9,
+    'quantizable_layer_type': ['Conv2D', 'Linear']}
+  print_model: True

configs/slim/quant/yolov3_darknet_qat.yml

+pretrain_weights: https://paddledet.bj.bcebos.com/models/yolov3_darknet53_270e_coco.pdparams
+slim: QAT
+
+QAT:
+  quant_config: {
+    'weight_quantize_type': 'channel_wise_abs_max', 'activation_quantize_type': 'moving_average_abs_max',
+    'weight_bits': 8, 'activation_bits': 8, 'dtype': 'int8', 'window_size': 10000, 'moving_rate': 0.9,
+    'quantizable_layer_type': ['Conv2D', 'Linear']}
+  print_model: True
+
+epoch: 50
+
+LearningRate:
+  base_lr: 0.0001
+  schedulers:
+  - !PiecewiseDecay
+    gamma: 0.1
+    milestones:
+    - 30
+    - 45
+  - !LinearWarmup
+    start_factor: 0.
+    steps: 1000
+
+OptimizerBuilder:
+  optimizer:
+    momentum: 0.9
+    type: Momentum
+  regularizer:
+    factor: 0.0005
+    type: L2

configs/slim/quant/yolov3_mobilenet_v1_qat.yml

 # Weights of yolov3_mobilenet_v1_coco
 pretrain_weights: https://paddledet.bj.bcebos.com/models/yolov3_mobilenet_v1_270e_coco.pdparams
-weight_type: resume
 slim: QAT
 
 QAT:

configs/slim/quant/yolov3_mobilenet_v3_qat.yml

 # Weights of yolov3_mobilenet_v3_coco
 pretrain_weights: https://paddledet.bj.bcebos.com/models/yolov3_mobilenet_v3_large_270e_coco.pdparams
-weight_type: resume
 slim: QAT
 
 QAT:
@@ -10,3 +9,16 @@ QAT:
     'weight_bits': 8, 'activation_bits': 8, 'dtype': 'int8', 'window_size': 10000, 'moving_rate': 0.9,
     'quantizable_layer_type': ['Conv2D', 'Linear']}
   print_model: True
+
+epoch: 30
+LearningRate:
+  base_lr: 0.0001
+  schedulers:
+  - !PiecewiseDecay
+    gamma: 0.1
+    milestones:
+    - 25
+    - 28
+  - !LinearWarmup
+    start_factor: 0.
+    steps: 2000

ppdet/modeling/heads/bbox_head.py

@@ -50,11 +50,13 @@ class TwoFCHead(nn.Layer):
             out_channel,
             weight_attr=paddle.ParamAttr(
                 initializer=XavierUniform(fan_out=fan)))
+        self.fc6.skip_quant = True
 
         self.fc7 = nn.Linear(
             out_channel,
             out_channel,
             weight_attr=paddle.ParamAttr(initializer=XavierUniform()))
+        self.fc7.skip_quant = True
 
     @classmethod
     def from_config(cls, cfg, input_shape):
@@ -199,12 +201,14 @@ class BBoxHead(nn.Layer):
             self.num_classes + 1,
             weight_attr=paddle.ParamAttr(initializer=Normal(
                 mean=0.0, std=0.01)))
+        self.bbox_score.skip_quant = True
 
         self.bbox_delta = nn.Linear(
             in_channel,
             4 * self.num_classes,
             weight_attr=paddle.ParamAttr(initializer=Normal(
                 mean=0.0, std=0.001)))
+        self.bbox_delta.skip_quant = True
         self.assigned_label = None
         self.assigned_rois = None
 

ppdet/modeling/heads/mask_head.py

@@ -65,20 +65,21 @@ class MaskFeat(nn.Layer):
                         norm_type=self.norm_type,
                         norm_name=conv_name + '_norm',
                         initializer=KaimingNormal(fan_in=fan_conv),
+                        skip_quant=True,
                         name=conv_name))
                 mask_conv.add_sublayer(conv_name + 'act', nn.ReLU())
         else:
             for i in range(self.num_convs):
                 conv_name = 'mask_inter_feat_{}'.format(i + 1)
-                mask_conv.add_sublayer(
-                    conv_name,
-                    nn.Conv2D(
+                conv = nn.Conv2D(
                     in_channels=in_channel if i == 0 else out_channel,
                     out_channels=out_channel,
                     kernel_size=3,
                     padding=1,
                     weight_attr=paddle.ParamAttr(
-                            initializer=KaimingNormal(fan_in=fan_conv))))
+                        initializer=KaimingNormal(fan_in=fan_conv)))
+                conv.skip_quant = True
+                mask_conv.add_sublayer(conv_name, conv)
                 mask_conv.add_sublayer(conv_name + 'act', nn.ReLU())
         mask_conv.add_sublayer(
             'conv5_mask',
@@ -146,6 +147,7 @@ class MaskHead(nn.Layer):
             kernel_size=1,
             weight_attr=paddle.ParamAttr(initializer=KaimingNormal(
                 fan_in=self.num_classes)))
+        self.mask_fcn_logits.skip_quant = True
 
     @classmethod
     def from_config(cls, cfg, input_shape):

ppdet/modeling/heads/yolo_head.py

@@ -47,16 +47,16 @@ class YOLOv3Head(nn.Layer):
             else:
                 num_filters = len(self.anchors[i]) * (self.num_classes + 5)
             name = 'yolo_output.{}'.format(i)
-            yolo_output = self.add_sublayer(
-                name,
-                nn.Conv2D(
+            conv = nn.Conv2D(
                 in_channels=128 * (2**self.num_outputs) // (2**i),
                 out_channels=num_filters,
                 kernel_size=1,
                 stride=1,
                 padding=0,
                 data_format=data_format,
-                    bias_attr=ParamAttr(regularizer=L2Decay(0.))))
+                bias_attr=ParamAttr(regularizer=L2Decay(0.)))
+            conv.skip_quant = True
+            yolo_output = self.add_sublayer(name, conv)
             self.yolo_outputs.append(yolo_output)
 
     def parse_anchor(self, anchors, anchor_masks):

ppdet/modeling/layers.py

@@ -52,6 +52,7 @@ class DeformableConvV2(nn.Layer):
                  bias_attr=None,
                  lr_scale=1,
                  regularizer=None,
+                 skip_quant=False,
                  name=None):
         super(DeformableConvV2, self).__init__()
         self.offset_channel = 2 * kernel_size**2
@@ -77,6 +78,8 @@ class DeformableConvV2(nn.Layer):
                 initializer=Constant(0.0),
                 name='{}._conv_offset.weight'.format(name)),
             bias_attr=offset_bias_attr)
+        if skip_quant:
+            self.conv_offset.skip_quant = True
 
         if bias_attr:
             # in FCOS-DCN head, specifically need learning_rate and regularizer
@@ -126,6 +129,7 @@ class ConvNormLayer(nn.Layer):
                  freeze_norm=False,
                  initializer=Normal(
                      mean=0., std=0.01),
+                 skip_quant=False,
                  name=None):
         super(ConvNormLayer, self).__init__()
         assert norm_type in ['bn', 'sync_bn', 'gn']
@@ -151,6 +155,8 @@ class ConvNormLayer(nn.Layer):
                     initializer=initializer,
                     learning_rate=1.),
                 bias_attr=bias_attr)
+            if skip_quant:
+                self.conv.skip_quant = True
         else:
             # in FCOS-DCN head, specifically need learning_rate and regularizer
             self.conv = DeformableConvV2(
@@ -167,6 +173,7 @@ class ConvNormLayer(nn.Layer):
                 bias_attr=True,
                 lr_scale=2.,
                 regularizer=L2Decay(norm_decay),
+                skip_quant=skip_quant,
                 name=name)
 
         norm_lr = 0. if freeze_norm else 1.

ppdet/modeling/proposal_generator/rpn_head.py

@@ -45,6 +45,7 @@ class RPNFeat(nn.Layer):
             padding=1,
             weight_attr=paddle.ParamAttr(initializer=Normal(
                 mean=0., std=0.01)))
+        self.rpn_conv.skip_quant = True
 
     def forward(self, feats):
         rpn_feats = []
@@ -100,6 +101,7 @@ class RPNHead(nn.Layer):
             padding=0,
             weight_attr=paddle.ParamAttr(initializer=Normal(
                 mean=0., std=0.01)))
+        self.rpn_rois_score.skip_quant = True
 
         # rpn roi bbox regression deltas
         self.rpn_rois_delta = nn.Conv2D(
@@ -109,6 +111,7 @@ class RPNHead(nn.Layer):
             padding=0,
             weight_attr=paddle.ParamAttr(initializer=Normal(
                 mean=0., std=0.01)))
+        self.rpn_rois_delta.skip_quant = True
 
     @classmethod
     def from_config(cls, cfg, input_shape):