CSP-Pedestrian-detection
CSP-Pedestrian-detection copied to clipboard
ligang-cs
A faster pytorch implementation of 'Center and Scale Prediction (CSP) for pedestrain detection (CVPR19)'.
Center and Scale Prediction (CSP) for pedestrian detection
Introduction
This is the unofficial pytorch implementation of High-level Semantic Feature Detection: A New Perspective for Pedestrian Detection. CSP is an effective and efficient method for pedestrian detector and achieves promising results on the CityPersons dataset. We implement CSP in pytorch based on previous works offical code (keras), unofficial code (pytorch). Compared with them, our codes have following features:
- Support Apex Mix Precision
- Support Distributed and Non-distributed training
- Support more backbones, such as ResNet, DLA-34, HRNet
We obtain much faster training/inference speed (3 hours for 120 epoches using two gpus) and comparable performance. We think CSP is a strong baseline for pedestrian detection, and it still has much room for improvement. We will continuously update this repo, and add some useful tricks (e.g. data augmentation) for better performance.
models
| Model | Reasonable | Heavy Occlusion | All | Training time | Link |
|---|---|---|---|---|---|
| ResNet-50 | 11.30 | 41.09 | 37.55 | ~5 hours | BaiduYun(code:v61g) |
| DLA-34 | 11.12 | 43.00 | 37.32 | ~3 hours | |
| HRNet-18 | 10.24 | 37.72 | 36.15 | ~11 hours | |
| HRNet-32 | 9.69 | 36.48 | 35.47 | ~13 hours | |
| HRNet-32 + SWA | 9.66 | 34.61 | 34.86 | BaiduYun(code:v61g) |
Note: Training time is evaluated in two 2080Ti GPUs for 120 epochs. We will further tune some hyperparameters (e.g. learning rate, batchsize) these days, then will release our models.
Get Start
Prerequisites
- Pytorch 1.2+
- Python 3.6+
- APEX (Install APEX following the offical instruction)
Installation
git clone [email protected]:ligang-cs/CSP-Pedestrain-detection.git
cd CSP-Pedestrian-detection/utils
make all
Data preparation
You need to download the CityPersons dataset.
Your directory tree should be look like this:
$root_path/
├── images
│ ├── train
│ └── val
├── annotations
│ ├── anno_train.mat
│ └── anno_val.mat
Train and test
Please specify the configuration file.
Distributed training
CUDA_VISIBLE_DEVICES=<gpus_ids> python -m torch.distributed.launch --nproc_per_node <gpus_number> trainval_distributed.py --work-dir <save_path>
Non-distributed training
CUDA_VISIBLE_DEVICES=<gpus_ids> python trainval.py --work-dir <save_path>
Test
CUDA_VISIBLE_DEVICES=<gpus_ids> python test.py --val-path <checkpoint_path> --json-out <results_path>
Contact
If you have any questions, please do not hesitate to contact Li Gang ([email protected]).
We also appreciate all contributions to improve this repo.
Acknowledgement
- HRNet codes are folked from HRNet offical codes
- DLA-34 codes are folked from here
Many thanks to them !
ligang-cs