https://github.com/hrnet/hrnet-fcos
High-resolution Networks for the Fully Convolutional One-Stage Object Detection (FCOS) algorithm
https://github.com/hrnet/hrnet-fcos
fcos hrnets mscoco object-detection
Last synced: 6 months ago
JSON representation
High-resolution Networks for the Fully Convolutional One-Stage Object Detection (FCOS) algorithm
- Host: GitHub
- URL: https://github.com/hrnet/hrnet-fcos
- Owner: HRNet
- License: other
- Created: 2019-06-28T06:00:25.000Z (over 6 years ago)
- Default Branch: master
- Last Pushed: 2019-10-21T10:59:48.000Z (almost 6 years ago)
- Last Synced: 2025-03-26T12:04:12.855Z (7 months ago)
- Topics: fcos, hrnets, mscoco, object-detection
- Language: Python
- Homepage:
- Size: 4.15 MB
- Stars: 125
- Watchers: 6
- Forks: 37
- Open Issues: 3
-
Metadata Files:
- Readme: README.md
- Contributing: CONTRIBUTING.md
- License: LICENSE
- Code of conduct: CODE_OF_CONDUCT.md
Awesome Lists containing this project
README
# High-resolution Networks for FCOS
## Introduction
This project contains the code of HRNet-FCOS, i.e., using [High-resolution Networks (HRNets)](https://arxiv.org/pdf/1904.04514.pdf) as the backbones for the [Fully Convolutional One-Stage Object Detection (FCOS)](https://arxiv.org/abs/1904.01355) algorithm, which achieves much better object detection performance compared with the ResNet-FCOS counterparts while keeping a similar computation complexity. For more projects using HRNets, please go to our [website](https://github.com/HRNet).
## Quick start
### Installation
Please check [INSTALL.md](INSTALL.md) for installation instructions.
You may also want to see the original [README.md](FCOS_README.md) of FCOS.
### Inference
The inference command line on coco minival split:
python tools/test_net.py \
--config-file configs/fcos/fcos_hrnet_w32_5l_2x.yaml \
MODEL.WEIGHT models/FCOS_hrnet_w32_5l_2x.pth \
TEST.IMS_PER_BATCH 8
Please note that:
1) If your model's name is different, please replace `models/FCOS_hrnet_w32_5l_2x.pth` with your own.
2) If you enounter out-of-memory error, please try to reduce `TEST.IMS_PER_BATCH` to 1.
3) If you want to evaluate a different model, please change `--config-file` to its config file (in [configs/fcos](configs/fcos)) and `MODEL.WEIGHT` to its weights file.
For your convenience, we provide the following trained models.
FCOS Model | Training mem (GB) | Multi-scale training | SyncBN| Testing time / im | # params |GFLOPs| AP (minival) | Link
--- |:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:
ResNet_50_5l_2x | 29.3 | No |No | 71ms |32.0M |190.0| 37.1 | [-]()
HRNet_W18_5l_2x | 54.4 | No |No | 72ms |17.5M |180.3| 37.7 | [model](https://1drv.ms/u/s!Av9x_1oQAAoqeRND03CfS4HBONM?e=wT0G0O)
HRNet_W18_5l_2x | 55.0 | Yes |Yes| 72ms |17.5M |180.3| 39.4 | [model](https://1drv.ms/u/s!Av9x_1oQAAoqgQlVjkAUDdV9Ej0A?e=9bh7VW)
||
ResNet_50_6l_2x | 58.2 | No |No | 98ms |32.7M |529.0| 37.1 | [-]()
HRNet_W18_6l_2x | 88.1 | No |No | 106ms |18.1M |515.1| 37.8 | [model](https://1drv.ms/u/s!Av9x_1oQAAoqeumBsKzXzZjE8Qs?e=fT1edk)
||
ResNet_101_5l_2x | 44.1 | Yes |No | 74ms |51.0M |261.2| 41.4 | [model](https://cloudstor.aarnet.edu.au/plus/s/vjL3L0AW7vnhRTo/download)
HRNet_W32_5l_2x | 78.9 | Yes |No | 87ms |37.3M |273.3| 41.9 | [model](https://1drv.ms/u/s!Av9x_1oQAAoqfPuN69wCHx26k0o?e=L7c5FX)
HRNet_W32_5l_2x | 80.1 | Yes |Yes| 87ms |37.3M |273.3| 42.5 | [model](https://1drv.ms/u/s!Av9x_1oQAAoqgQHJWCW0-u0zOgzi?e=yrtKUt)
||
ResNet_101_6l_2x | 71.0 | Yes |No | 121ms |51.6M |601.0| 41.5 | [model](https://1drv.ms/u/s!Av9x_1oQAAoqe7UH3Bh-kB8JuKA?e=EF9K0B)
HRNet_W32_6l_2x | 108.6| Yes |No | 125ms |37.9M |608.0| 42.1 | [model](https://1drv.ms/u/s!Av9x_1oQAAoqfZn3Xt2CrKUI2rk?e=ZdJSPG)
HRNet_W32_6l_2x | 109.9| Yes |Yes| 125ms |37.9M |608.0| 42.9 | [model](https://1drv.ms/u/s!Av9x_1oQAAoqf6I0plglrSXPuys?e=lzdDwu)
||
HRNet_W40_6l_3x | 128.0| Yes |No | 142ms |54.1M |682.9| 42.6 | [model](https://1drv.ms/u/s!Av9x_1oQAAoqfu-2x6aOIsGxSsg?e=OBbs5Z)
[1] *1x, 2x and 3x mean the model is trained for 90K, 180K and 270k iterations, respectively.*\
[2] *5l and 6l denote that we use feature pyramid with 5 levels and 6 levels, respectively.*\
[3] *We provide model trained with Synchronous Batch Normalization (SyncBN).*\
[4] *We report total training memory footprint on all GPUs instead of the memory footprint per GPU as in maskrcnn-benchmark.*\
[5] *The inference speed of HRNet can get improved if the branches in the HRNet model can run in parallel.*\
[6] *All results are obtained with a single model and without any test time data augmentation.*
### Training
The following command line will trains a fcos_hrnet_w32_5l_2x model on 8 GPUs with Synchronous Stochastic Gradient Descent (SGD):
python -m torch.distributed.launch \
--nproc_per_node=8 \
--master_port=$((RANDOM + 10000)) \
tools/train_net.py \
--config-file configs/fcos/fcos_hrnet_w32_5l_2x.yaml \
MODEL.WEIGHT hrnetv2_w32_imagenet_pretrained.pth \
MODEL.SYNCBN False \
DATALOADER.NUM_WORKERS 4 \
OUTPUT_DIR training_dir/fcos_hrnet_w32_5l_2x
Note that:
1) If you want to use fewer GPUs, please change `--nproc_per_node` to the number of GPUs. No other settings need to be changed. The total batch size does not depends on `nproc_per_node`. If you want to change the total batch size, please change `SOLVER.IMS_PER_BATCH` in [configs/fcos/fcos_hrnet_w32_5l_2x.yaml](configs/fcos/fcos_hrnet_w32_5l_2x.yaml).
2) If you want to use Synchronous Batch-Normalization (SyncBN), please change `MODEL.SYNCBN` to `True`. Note that this will lead to ~2x slower training speed when training on mulitple machines. You also need to fix the image padding size when using SyncBN, see [here](maskrcnn_benchmark/structures/image_list.py#L62).
3) The imagenet pre-trained model can be found [here](https://github.com/HRNet/HRNet-Image-Classification#imagenet-pretrained-models).
4) The models will be saved into `OUTPUT_DIR`.
5) If you want to train FCOS on your own dataset, please follow this instruction [#54](https://github.com/tianzhi0549/FCOS/issues/54#issuecomment-497558687).
### Contributing to the project
Any pull requests or issues are welcome.
### Citations
Please consider citing the following papers in your publications if the project helps your research.
```
@article{sun2019deep,
title={Deep High-Resolution Representation Learning for Human Pose Estimation},
author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
journal={arXiv preprint arXiv:1902.09212},
year={2019}
}
@article{tian2019fcos,
title = {{FCOS}: Fully Convolutional One-Stage Object Detection},
author = {Tian, Zhi and Shen, Chunhua and Chen, Hao and He, Tong},
journal = {arXiv preprint arXiv:1904.01355},
year = {2019}
}
```
### License
For academic use, this project is licensed under the 2-clause BSD License - see the LICENSE file for details. For commercial use, please contact the authors.