Reversible Column Networks
We propose a new neural network design paradigm Reversible Column Network (RevCol). The main body of RevCol is composed of multiple copies of subnetworks, named columns respectively, between which multi-level reversible connections are employed. Such architectural scheme attributes RevCol very different behavior from conventional networks: during forward propagation, features in RevCol are learned to be gradually disentangled when passing through each column, whose total information is maintained rather than compressed or discarded as other network does. Our experiments suggest that CNN-style RevCol models can achieve very competitive performances on multiple computer vision tasks such as image classification, object detection and semantic segmentation, especially with large parameter budget and large dataset. For example, after ImageNet-22K pre-training, RevCol-XL obtains 88.2% ImageNet-1K accuracy. Given more pre-training data, our largest model RevCol-H reaches 90.0% on ImageNet-1K, 63.8% APbox on COCO detection minival set, 61.0% mIoU on ADE20k segmentation. To our knowledge, it is the best COCO detection and ADE20k segmentation result among pure (static) CNN models. Moreover, as a general macro architecture fashion, RevCol can also be introduced into transformers or other neural networks, which is demonstrated to improve the performances in both computer vision and NLP tasks. We release code and models at https://github.com/megvii-research/RevCol
PDF AbstractCode
Datasets
ImageNet
MS COCO
ADE20K
Objects365
Results from the Paper
Ranked #8 on
Semantic Segmentation
on ADE20K
(using extra training data)
| Task | Dataset | Model | Metric Name | Metric Value | Global Rank | Uses Extra Training Data |
Benchmark |
|---|---|---|---|---|---|---|---|
| Semantic Segmentation | ADE20K | RevCol-H (Mask2Former) | Validation mIoU | 61.0 | # 8 | ||
| Object Detection | COCO minival | RevCol-H(DINO) | box AP | 63.8 | # 10 | ||
| Object Detection | COCO test-dev | RevCol-H(DINO) | box mAP | 63.8 | # 12 | ||
| Image Classification | ImageNet | RevCol-H | Top 1 Accuracy | 90.0% | # 19 | ||
| Number of params | 2158M | # 968 |
