1. 量化#
1.1 图象分类#
数据集：ImageNet1000类
评价指标：Top-1/Top-5准确率
| Model | FP32 | 离线量化 | 量化训练 |
|---|---|---|---|
| MobileNetV1 FP32 | 70.99%/89.68% | xx%/xx% | xx%/xx% |
| MobileNetV2 FP32 | 72.15%/90.65% | ||
| ResNet50 FP32 | 76.50%/93.00% |
量化前后，模型大小的变化对比如下：
| Model | FP32 | 离线量化 | 量化训练 |
|---|---|---|---|
| MobileNetV1 | 17M | xxM | xxM |
| MobileNetV2 | xxM | ||
| ResNet50 | 99M |
1.2 目标检测#
数据集：COCO 2017
| Model | 输入尺寸 | Image/GPU | FP32 BoxAP | 离线量化 BoxAP | 量化训练 BoxAP |
|---|---|---|---|---|---|
| MobileNet-V1-YOLOv3 | 608 | 8 | 29.3 | xx | xx |
| MobileNet-V1-YOLOv3 | 416 | ||||
| MobileNet-V1-YOLOv3 | 320 | ||||
| R50-dcn-YOLOv3 | 608 | 41.4 | |||
| R50-dcn-YOLOv3 | 416 | ||||
| R50-dcn-YOLOv3 | 320 |
量化前后，模型大小的变化对比如下：
| Model | FP32 | 离线量化 | 量化训练 |
|---|---|---|---|
| MobileNet-V1-YOLOv3 | xxM | xxM | xxM |
| R50-dcn-YOLOv3 | xxM |
#
数据集：WIDER-FACE
评价指标：Easy/Medium/Hard mAP
| Model | 输入尺寸 | Image/GPU | FP32 | 离线量化 | 量化训练 |
|---|---|---|---|---|---|
| BlazeFace | 640 | 8 | 0.915/0.892/0.797 | xx/xx/xx | xx/xx/xx |
| BlazeFace-Lite | 640 | 0.909/0.885/0.781 | |||
| BlazeFace-NAS | 640 | 0.837/0.807/0.658 |
量化前后，模型大小的变化对比如下：
| Model | FP32 | 离线量化 | 量化训练 |
|---|---|---|---|
| BlazeFace | xxM | xxM | xxM |
| BlazeFace-Lite | xxM | ||
| BlazeFace-NAS | xxM |
#
1.3 图像分割#
数据集：Cityscapes
| Model | FP32 mIoU | 离线量化 mIoU | 量化训练 mIoU |
|---|---|---|---|
| DeepLabv3+/MobileNetv1 | 63.26 | xx | xx |
| DeepLabv3+/MobileNetv2 | 69.81 |
量化前后，模型大小的变化对比如下：
| Model | FP32 | 离线量化 | 量化训练 |
|---|---|---|---|
| DeepLabv3+/MobileNetv1 | xxM | xxM | xxM |
| DeepLabv3+/MobileNetv2 | xxM |
#
2. 剪枝#
2.1 图像分类#
数据集：ImageNet1000类
| Model | Top-1/Top-5 |
|---|---|
| MobileNetV1 | 70.99%/89.68% |
| MobileNetV1 uniform -50% | |
| MobileNetV1 sensitive -xx% | |
| MobileNetV2 | 72.15%/90.65% |
| MobileNetV2 uniform -50% | |
| MobileNetV2 sensitive -xx% | |
| ResNet34 | 74.57%/92.14% |
| ResNet34 uniform -50% | |
| ResNet34 auto -50% |
剪枝前后，模型大小和计算量的变化对比如下：
| Model | baseline FLOPs | baseline size | 剪枝后 FlOPs | 剪枝后 size |
|---|---|---|---|---|
| MobileNetV1 | xx | xx | xx | xx |
| MobileNetV2 | xx | |||
| ResNet34 | xx |
#
2.2 目标检测#
数据集：Pasacl VOC & COCO 2017
| Model | 数据集 | 输入尺寸 | Image/GPU | baseline mAP | 敏感度剪枝 mAP |
|---|---|---|---|---|---|
| MobileNet-V1-YOLOv3 | Pasacl VOC | 608 | 8 | 76.2 | 77.59 (-50%) |
| MobileNet-V1-YOLOv3 | Pasacl VOC | 416 | 76.7 | xx | |
| MobileNet-V1-YOLOv3 | Pasacl VOC | 320 | 75.2 | xx | |
| MobileNet-V1-YOLOv3 | COCO | 608 | 29.3 | 29.56 (-20%) | |
| MobileNet-V1-YOLOv3 | COCO | 416 | 29.3 | xx | |
| MobileNet-V1-YOLOv3 | COCO | 320 | 27.1 | xx | |
| R50-dcn-YOLOv3 | COCO | 608 | 41.4 | 37.8 (-30%) | |
| R50-dcn-YOLOv3 | COCO | 416 | |||
| R50-dcn-YOLOv3 | COCO | 320 |
剪枝前后，模型大小和计算量的变化对比如下：
| Model | baseline FLOPs | baseline size | 剪枝后 FlOPs | 剪枝后 size |
|---|---|---|---|---|
| MobileNet-V1-YOLOv3-VOC | xx | xxM | xx | xxM |
| MobileNet-V1-YOLOv3-COCO | xx | |||
| R50-dcn-YOLOv3-COCO | xx |
2.3 图像分割#
数据集：Cityscapes
| Model | Baseline mIoU | xx剪枝 mIoU |
|---|---|---|
| DeepLabv3+/MobileNetv2 | 69.81 | xx |
剪枝前后，模型大小和计算量的变化对比如下：
| Model | baseline FLOPs | baseline size | 剪枝后 FlOPs | 剪枝后 size |
|---|---|---|---|---|
| DeepLabv3+/MobileNetv2 | xx | xxM | xx | xxM |
3. 蒸馏#
3.1 图象分类#
数据集：ImageNet1000类
评价指标：Top-1/Top-5准确率
| Model | baseline | 蒸馏后 |
|---|---|---|
| MobileNetV1 | 70.99%/89.68% | 72.79%/90.69% (teacher: ResNet50_vd1) |
| MobileNetV2 | 72.15%/90.65% | 74.30%/91.52% (teacher: ResNet50_vd) |
| ResNet50 | 76.50%/93.00% | 77.40%/93.48% (teacher: ResNet1012) |
Note
[1]：ResNet50_vd预训练模型Top-1/Top-5准确率分别为79.12%/94.44%
带_vd后缀代表开启了Mixup训练，Mixup相关介绍参考mixup: Beyond Empirical Risk Minimization
3.2 目标检测#
数据集：Pasacl VOC & COCO 2017
| Model | 数据集 | 输入尺寸 | Image/GPU | baseline | 蒸馏后 mAP |
|---|---|---|---|---|---|
| MobileNet-V1-YOLOv3 | Pasacl VOC | 640 | 16 | 76.2 | 79.0 (teacher: ResNet34-YOLOv3-VOC3) |
| MobileNet-V1-YOLOv3 | Pasacl VOC | 416 | 76.7 | 78.2 | |
| MobileNet-V1-YOLOv3 | Pasacl VOC | 320 | 75.2 | 75.5 | |
| MobileNet-V1-YOLOv3 | COCO | 640 | 29.3 | 31.0 (teacher: ResNet34-YOLOv3-COCO4) | |
| MobileNet-V1-YOLOv3 | COCO | 416 | |||
| MobileNet-V1-YOLOv3 | COCO | 320 |