Tong He, Zhi Zhang, Hang Zhang, Zhongyue Zhang, Junyuan Xie, Mu Li
Much of the recent progress made in image classification research can be credited to training procedure refinements, such as changes in data augmentations and optimization methods. In the literature, however, most refinements are either briefly mentioned as implementation details or only visible in source code. In this paper, we will examine a collection of such refinements and empirically evaluate their impact on the final model accuracy through ablation study. We will show that, by combining these refinements together, we are able to improve various CNN models significantly. For example, we raise ResNet-50's top-1 validation accuracy from 75.3% to 79.29% on ImageNet. We will also demonstrate that improvement on image classification accuracy leads to better transfer learning performance in other application domains such as object detection and semantic segmentation.
| Task | Dataset | Metric | Value | Model |
|---|---|---|---|---|
| Domain Adaptation | VizWiz-Classification | Accuracy - All Images | 39.7 | ResNet-26-D |
| Domain Adaptation | VizWiz-Classification | Accuracy - Clean Images | 43.5 | ResNet-26-D |
| Domain Adaptation | VizWiz-Classification | Accuracy - Corrupted Images | 35.8 | ResNet-26-D |
| Domain Generalization | VizWiz-Classification | Accuracy - All Images | 39.7 | ResNet-26-D |
| Domain Generalization | VizWiz-Classification | Accuracy - Clean Images | 43.5 | ResNet-26-D |
| Domain Generalization | VizWiz-Classification | Accuracy - Corrupted Images | 35.8 | ResNet-26-D |