Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/donnyyou/torchcv
TorchCV: A PyTorch-Based Framework for Deep Learning in Computer Vision
https://github.com/donnyyou/torchcv
Last synced: 5 days ago
JSON representation
TorchCV: A PyTorch-Based Framework for Deep Learning in Computer Vision
- Host: GitHub
- URL: https://github.com/donnyyou/torchcv
- Owner: donnyyou
- License: apache-2.0
- Created: 2018-10-19T03:38:47.000Z (about 6 years ago)
- Default Branch: master
- Last Pushed: 2020-11-19T05:40:57.000Z (about 4 years ago)
- Last Synced: 2024-11-21T06:02:23.482Z (21 days ago)
- Language: Shell
- Homepage: https://pytorchcv.com
- Size: 28.8 MB
- Stars: 2,250
- Watchers: 69
- Forks: 378
- Open Issues: 42
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
- Awesome-pytorch-list-CNVersion - TorchCV
- awesome-list - TorchCV - A PyTorch-Based Framework for Deep Learning in Computer Vision. (Computer Vision / General Purpose CV)
- Awesome-pytorch-list - TorchCV - Based Framework for Deep Learning in Computer Vision. (Pytorch & related libraries / CV:)
README
# TorchCV: A PyTorch-Based Framework for Deep Learning in Computer Vision
```
@misc{you2019torchcv,
author = {Ansheng You and Xiangtai Li and Zhen Zhu and Yunhai Tong},
title = {TorchCV: A PyTorch-Based Framework for Deep Learning in Computer Vision},
howpublished = {\url{https://github.com/donnyyou/torchcv}},
year = {2019}
}
```
This repository provides source code for most deep learning based cv problems. We'll do our best to keep this repository up-to-date. If you do find a problem about this repository, please raise an issue or submit a pull request.
```diff
- Semantic Flow for Fast and Accurate Scene Parsing
- Code and models: https://github.com/lxtGH/SFSegNets
```
## Implemented Papers
- [Image Classification](https://github.com/youansheng/torchcv/tree/master/runner/cls)
- VGG: Very Deep Convolutional Networks for Large-Scale Image Recognition
- ResNet: Deep Residual Learning for Image Recognition
- DenseNet: Densely Connected Convolutional Networks
- ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices
- ShuffleNet V2: Practical Guidelines for Ecient CNN Architecture Design
- Partial Order Pruning: for Best Speed/Accuracy Trade-off in Neural Architecture Search
- [Semantic Segmentation](https://github.com/youansheng/torchcv/tree/master/runner/seg)
- DeepLabV3: Rethinking Atrous Convolution for Semantic Image Segmentation
- PSPNet: Pyramid Scene Parsing Network
- DenseASPP: DenseASPP for Semantic Segmentation in Street Scenes
- Asymmetric Non-local Neural Networks for Semantic Segmentation
- Semantic Flow for Fast and Accurate Scene Parsing
- [Object Detection](https://github.com/youansheng/torchcv/tree/master/runner/det)
- SSD: Single Shot MultiBox Detector
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
- YOLOv3: An Incremental Improvement
- FPN: Feature Pyramid Networks for Object Detection
- [Pose Estimation](https://github.com/youansheng/torchcv/tree/master/runner/pose)
- CPM: Convolutional Pose Machines
- OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields
- [Instance Segmentation](https://github.com/youansheng/torchcv/tree/master/runner/seg)
- Mask R-CNN
- [Generative Adversarial Networks](https://github.com/youansheng/torchcv/tree/master/runner/gan)
- Pix2pix: Image-to-Image Translation with Conditional Adversarial Nets
- CycleGAN: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks.
## QuickStart with TorchCV
Now only support Python3.x, pytorch 1.3.
```bash
pip3 install -r requirements.txt
cd lib/exts
sh make.sh
```
## Performances with TorchCV
All the performances showed below fully reimplemented the papers' results.
#### Image Classification
- ImageNet (Center Crop Test): 224x224
| Model | Train | Test | Top-1 | Top-5 | BS | Iters | Scripts |
|:--------|:---------|:------|:------|:------|:------|:------|:------|
| ResNet50 | train | val | 77.54 | 93.59 | 512 | 30W | [ResNet50](https://github.com/youansheng/torchcv/blob/master/scripts/cls/imagenet/run_ic_res50_imagenet_cls.sh) |
| ResNet101 | train | val | 78.94 | 94.56 | 512 | 30W | [ResNet101](https://github.com/youansheng/torchcv/blob/master/scripts/cls/imagenet/run_ic_res101_imagenet_cls.sh) |
| ShuffleNetV2x0.5 | train | val | 60.90 | 82.54 | 1024 | 40W | [ShuffleNetV2x0.5](https://github.com/youansheng/torchcv/blob/master/scripts/cls/imagenet/run_ic_shufflenetv2x0.5_imagenet_cls.sh) |
| ShuffleNetV2x1.0 | train | val | 69.71 | 88.91 | 1024 | 40W | [ShuffleNetV2x1.0](https://github.com/youansheng/torchcv/blob/master/scripts/cls/imagenet/run_ic_shufflenetv2x1.0_imagenet_cls.sh) |
| DFNetV1 | train | val | 70.99 | 89.68 | 1024 | 40W | [DFNetV1](https://github.com/youansheng/torchcv/blob/master/scripts/cls/imagenet/run_ic_dfnetv1_imagenet_cls.sh) |
| DFNetV2 | train | val | 74.22 | 91.61 | 1024 | 40W | [DFNetV2](https://github.com/youansheng/torchcv/blob/master/scripts/cls/imagenet/run_ic_dfnetv2_imagenet_cls.sh) |
#### Semantic Segmentation
- Cityscapes (Single Scale Whole Image Test): Base LR 0.01, Crop Size 769
| Model | Backbone | Train | Test | mIOU | BS | Iters | Scripts |
|:--------|:---------|:------|:------|:------|:------|:------|:------|
| [PSPNet]() | [3x3-Res101](https://drive.google.com/open?id=1bUzCKazlh8ElGVYWlABBAb0b0uIqFgtR) | train | val | 78.20 | 8 | 4W | [PSPNet](https://github.com/youansheng/torchcv/blob/master/scripts/seg/cityscapes/run_fs_pspnet_cityscapes_seg.sh) |
| [DeepLabV3]() | [3x3-Res101](https://drive.google.com/open?id=1bUzCKazlh8ElGVYWlABBAb0b0uIqFgtR) | train | val | 79.13 | 8 | 4W | [DeepLabV3](https://github.com/youansheng/torchcv/blob/master/scripts/seg/cityscapes/run_fs_deeplabv3_cityscapes_seg.sh) |
- ADE20K (Single Scale Whole Image Test): Base LR 0.02, Crop Size 520
| Model | Backbone | Train | Test | mIOU | PixelACC | BS | Iters | Scripts |
|:--------|:---------|:------|:------|:------|:------|:------|:------|:------|
| [PSPNet]() | [3x3-Res50](https://drive.google.com/open?id=1zPQLFd9c1yHfkQn5CWBCcEKmjEEqxsWx) | train | val | 41.52 | 80.09 | 16 | 15W | [PSPNet](https://github.com/youansheng/torchcv/blob/master/scripts/seg/ade20k/run_fs_res50_pspnet_ade20k_seg.sh) |
| [DeepLabv3]() | [3x3-Res50](https://drive.google.com/open?id=1zPQLFd9c1yHfkQn5CWBCcEKmjEEqxsWx) | train | val | 42.16 | 80.36 | 16 | 15W | [DeepLabV3](https://github.com/youansheng/torchcv/blob/master/scripts/seg/ade20k/run_fs_res50_deeplabv3_ade20k_seg.sh) |
| [PSPNet]() | [3x3-Res101](https://drive.google.com/open?id=1bUzCKazlh8ElGVYWlABBAb0b0uIqFgtR) | train | val | 43.60 | 81.30 | 16 | 15W | [PSPNet](https://github.com/youansheng/torchcv/blob/master/scripts/seg/ade20k/run_fs_res101_pspnet_ade20k_seg.sh) |
| [DeepLabv3]() | [3x3-Res101](https://drive.google.com/open?id=1bUzCKazlh8ElGVYWlABBAb0b0uIqFgtR) | train | val | 44.13 | 81.42 | 16 | 15W | [DeepLabV3](https://github.com/youansheng/torchcv/blob/master/scripts/seg/ade20k/run_fs_res101_deeplabv3_ade20k_seg.sh) |
#### Object Detection
- Pascal VOC2007/2012 (Single Scale Test): 20 Classes
| Model | Backbone | Train | Test | mAP | BS | Epochs | Scripts |
|:--------|:---------|:------|:------|:------|:------|:------|:------|
| [SSD300](https://drive.google.com/open?id=15J5blVyZq7lqCePh-Q8S2pxim3-f_8LP) | [VGG16](https://drive.google.com/open?id=1nM0UwmqR4lIHzmRWvs71jfP_gAekjuKy) | 07+12_trainval | 07_test | 0.786 | 32 | 235 | [SSD300](https://github.com/youansheng/torchcv/blob/master/scripts/det/voc/run_ssd300_vgg16_voc_det.sh) |
| [SSD512](https://drive.google.com/open?id=1RF5gnqfiyz-EcSFU1OSK7tNuX_VRObVW) | [VGG16](https://drive.google.com/open?id=1nM0UwmqR4lIHzmRWvs71jfP_gAekjuKy) | 07+12_trainval | 07_test | 0.808 | 32 | 235 | [SSD512](https://github.com/youansheng/torchcv/blob/master/scripts/det/voc/run_ssd512_vgg16_voc_det.sh) |
| [Faster R-CNN](https://drive.google.com/open?id=15SfklRiI1McVWEq9EAceznK-9sxXSQR4) | [VGG16](https://drive.google.com/open?id=1ZL9SS9KRzsDQhMe8kyPQ1LHA60wx_Vcj) | 07_trainval | 07_test | 0.706 | 1 | 15 | [Faster R-CNN](https://github.com/youansheng/torchcv/blob/master/scripts/det/voc/run_fr_vgg16_voc_det.sh) |
#### Pose Estimation
- OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields
#### Instance Segmentation
- Mask R-CNN
#### Generative Adversarial Networks
- Pix2pix
- CycleGAN
## DataSets with TorchCV
TorchCV has defined the dataset format of all the tasks which you could check in the subdirs of [data](https://github.com/youansheng/torchcv/tree/master/data). Following is an example dataset directory trees for training semantic segmentation. You could preprocess the open datasets with the scripts in folder [data/seg/preprocess](https://github.com/youansheng/torchcv/tree/master/data/seg/preprocess)
```
Dataset
train
image
00001.jpg/png
00002.jpg/png
...
label
00001.png
00002.png
...
val
image
00001.jpg/png
00002.jpg/png
...
label
00001.png
00002.png
...
```
## Commands with TorchCV
Take PSPNet as an example. ("tag" could be any string, include an empty one.)
- Training
```bash
cd scripts/seg/cityscapes/
bash run_fs_pspnet_cityscapes_seg.sh train tag
```
- Resume Training
```bash
cd scripts/seg/cityscapes/
bash run_fs_pspnet_cityscapes_seg.sh train tag
```
- Validate
```bash
cd scripts/seg/cityscapes/
bash run_fs_pspnet_cityscapes_seg.sh val tag
```
- Testing:
```bash
cd scripts/seg/cityscapes/
bash run_fs_pspnet_cityscapes_seg.sh test tag
```
## Demos with TorchCV
Example output of VGG19-OpenPose
Example output of VGG19-OpenPose