LarryTsai commited on
Commit
2b362b3
1 Parent(s): b9425fd

Upload README.md

Browse files
Files changed (1) hide show
  1. README.md +79 -3
README.md CHANGED
@@ -1,3 +1,79 @@
1
- ---
2
- license: apache-2.0
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Reversible Column Networks
2
+ This repo is the official implementation of:
3
+
4
+ ### [Reversible Column Networks](https://arxiv.org/abs/2212.11696)
5
+ [Yuxuan Cai](https://nightsnack.github.io), [Yizhuang Zhou](https://scholar.google.com/citations?user=VRSGDDEAAAAJ), [Qi Han](https://hanqer.github.io), Jianjian Sun, Xiangwen Kong, Jun Li, [Xiangyu Zhang](https://scholar.google.com/citations?user=yuB-cfoAAAAJ) \
6
+ [MEGVII Technology](https://en.megvii.com)\
7
+ International Conference on Learning Representations (ICLR) 2023\
8
+ [\[arxiv\]](https://arxiv.org/abs/2212.11696)
9
+
10
+
11
+
12
+ ## Updates
13
+ ***2/10/2023***\
14
+ RevCol model weights released.
15
+
16
+ ***1/21/2023***\
17
+ RevCol was accepted by ICLR 2023!
18
+
19
+ ***12/23/2022***\
20
+ Initial commits: codes for ImageNet-1k and ImageNet-22k classification are released.
21
+
22
+
23
+ ## To Do List
24
+
25
+
26
+ - [x] ImageNet-1K and 22k Training Code
27
+ - [x] ImageNet-1K and 22k Model Weights
28
+ - [ ] Cascade Mask R-CNN COCO Object Detection Code & Model Weights
29
+ - [ ] ADE20k Semantic Segmentation Code & Model Weights
30
+
31
+
32
+ ## Introduction
33
+ RevCol is composed of multiple copies of subnetworks, named columns respectively, between which multi-level reversible connections are employed. RevCol coud serves as a foundation model backbone for various tasks in computer vision including classification, detection and segmentation.
34
+
35
+ <p align="center">
36
+ <img src="figures/title.png" width=100% height=100%
37
+ class="center">
38
+ </p>
39
+
40
+ ## Main Results on ImageNet with Pre-trained Models
41
+
42
+ | name | pretrain | resolution | #params |FLOPs | acc@1 | pretrained model | finetuned model |
43
+ |:---------------------:| :---: | :---: | :---: | :---: | :---: | :---: | :---: |
44
+ | RevCol-T | ImageNet-1K | 224x224 | 30M | 4.5G | 82.2 | [baidu](https://pan.baidu.com/s/1iGsbdmFcDpwviCHaajeUnA?pwd=h4tj)/[github](https://github.com/megvii-research/RevCol/releases/download/checkpoint/revcol_tiny_1k.pth) | - |
45
+ | RevCol-S | ImageNet-1K | 224x224 | 60M | 9.0G | 83.5 | [baidu](https://pan.baidu.com/s/1hpHfdFrTZIPB5NTwqDMLag?pwd=mxuk)/[github](https://github.com/megvii-research/RevCol/releases/download/checkpoint/revcol_small_1k.pth) | - |
46
+ | RevCol-B | ImageNet-1K | 224x224 | 138M | 16.6G | 84.1 | [baidu](https://pan.baidu.com/s/16XIJ1n8pXPD2cXwnFX6b9w?pwd=j6x9)/[github](https://github.com/megvii-research/RevCol/releases/download/checkpoint/revcol_base_1k.pth) | - |
47
+ | RevCol-B<sup>\*</sup> | ImageNet-22K | 224x224 | 138M | 16.6G | 85.6 |[baidu](https://pan.baidu.com/s/1l8zOFifgC8fZtBpHK2ZQHg?pwd=rh58)/[github](https://github.com/megvii-research/RevCol/releases/download/checkpoint/revcol_base_22k.pth)| [baidu](https://pan.baidu.com/s/1HqhDXL6OIQdn1LeM2pewYQ?pwd=1bp3)/[github](https://github.com/megvii-research/RevCol/releases/download/checkpoint/revcol_base_22k_1kft_224.pth)|
48
+ | RevCol-B<sup>\*</sup> | ImageNet-22K | 384x384 | 138M | 48.9G | 86.7 |[baidu](https://pan.baidu.com/s/1l8zOFifgC8fZtBpHK2ZQHg?pwd=rh58)/[github](https://github.com/megvii-research/RevCol/releases/download/checkpoint/revcol_base_22k.pth)| [baidu](https://pan.baidu.com/s/18G0zAUygKgu58s2AjCBpsw?pwd=rv86)/[github](https://github.com/megvii-research/RevCol/releases/download/checkpoint/revcol_base_22k_1kft_384.pth)|
49
+ | RevCol-L<sup>\*</sup> | ImageNet-22K | 224x224 | 273M | 39G | 86.6 |[baidu](https://pan.baidu.com/s/1ueKqh3lFAAgC-vVU34ChYA?pwd=qv5m)/[github](https://github.com/megvii-research/RevCol/releases/download/checkpoint/revcol_large_22k.pth)| [baidu](https://pan.baidu.com/s/1CsWmcPcwieMzXE8pVmHh7w?pwd=qd9n)/[github](https://github.com/megvii-research/RevCol/releases/download/checkpoint/revcol_large_22k_1kft_224.pth)|
50
+ | RevCol-L<sup>\*</sup> | ImageNet-22K | 384x384 | 273M | 116G | 87.6 |[baidu](https://pan.baidu.com/s/1ueKqh3lFAAgC-vVU34ChYA?pwd=qv5m)/[github](https://github.com/megvii-research/RevCol/releases/download/checkpoint/revcol_large_22k.pth)| [baidu](https://pan.baidu.com/s/1VmCE3W3Xw6-Lo4rWrj9Xzg?pwd=x69r)/[github](https://github.com/megvii-research/RevCol/releases/download/checkpoint/revcol_large_22k_1kft_384.pth)|
51
+
52
+ ## Getting Started
53
+ Please refer to [INSTRUCTIONS.md](INSTRUCTIONS.md) for setting up, training and evaluation details.
54
+
55
+
56
+ ## Acknowledgement
57
+ This repo was inspired by several open source projects. We are grateful for these excellent projects and list them as follows:
58
+ - [timm](https://github.com/rwightman/pytorch-image-models)
59
+ - [Swin Transformer](https://github.com/microsoft/Swin-Transformer)
60
+ - [ConvNeXt](https://github.com/facebookresearch/ConvNeXt)
61
+ - [beit](https://github.com/microsoft/unilm/tree/master/beit)
62
+
63
+ ## License
64
+ RevCol is released under the [Apache 2.0 license](LICENSE).
65
+
66
+ ## Contact Us
67
+ If you have any questions about this repo or the original paper, please contact Yuxuan at [email protected].
68
+
69
+
70
+ ## Citation
71
+ ```
72
+ @inproceedings{cai2022reversible,
73
+ title={Reversible Column Networks},
74
+ author={Cai, Yuxuan and Zhou, Yizhuang and Han, Qi and Sun, Jianjian and Kong, Xiangwen and Li, Jun and Zhang, Xiangyu},
75
+ booktitle={International Conference on Learning Representations},
76
+ year={2023},
77
+ url={https://openreview.net/forum?id=Oc2vlWU0jFY}
78
+ }
79
+ ```