An improved version of refinedet network, modify the backbone network, and provide resnet50 compression network and resnest50 method
A higher performance PyTorch implementation of Single-Shot Refinement Neural Network for Object Detection. The official and original Caffe code can be found here.
###简介 本实验我没有在VOC上进行测试对比,针对于实际项目实验发现:resnet50-86-refinedet的检测网络,耗时方面较原本vgg时间减少了一半,resnest50的性能较resnet50也有很大的提升。
Arch | Paper | Caffe Version | Our PyTorch Version |
---|---|---|---|
RefineDet320 | 80.0% | 79.52% | 79.81% |
RefineDet512 | 81.8% | 81.85% | 80.50% |
- Install PyTorch by selecting your environment on the website and running the appropriate command.
- Note: You should use at least PyTorch1.0.0
- Clone this repository.
- Note: We currently only support Python 3+.
- Then download the dataset by following the instructions below.
- We now support Visdom for real-time loss visualization during training!
- To use Visdom in the browser:
# First install Python server and client pip install visdom # Start the server (probably in a screen or tmux) python -m visdom.server
- Then (during training) navigate to http://localhost:8097/ (see the Train section below for training details).
- Note: For training, we currently support VOC and COCO, and aim to add ImageNet support soon.
To make things easy, we provide bash scripts to handle the dataset downloads and setup for you. We also provide simple dataset loaders that inherit torch.utils.data.Dataset
, making them fully compatible with the torchvision.datasets
API.
Microsoft COCO: Common Objects in Context
# specify a directory for dataset to be downloaded into, else default is ~/data/
sh data/scripts/COCO2014.sh
PASCAL VOC: Visual Object Classes
# specify a directory for dataset to be downloaded into, else default is ~/data/
sh data/scripts/VOC2007.sh # <directory>
# specify a directory for dataset to be downloaded into, else default is ~/data/
sh data/scripts/VOC2012.sh # <directory>
-
First download the fc-reduced Resnest PyTorch base network weights at: https://hangzh.s3.amazonaws.com/encoding/models/resnest50-528c19ca.pth
-
and, I have downloaded the file in the
resnest-refinedet/weights
dir:resnest50;秘钥4a7k -
To train resnest50-refinedet . You can manually change them as you want.
python train_refinedet.py
- Observe the training loss
python -m visdom.server
Open the web input website: http://localhost:8097/
- Note:
- For training, an NVIDIA GPU is strongly recommended for speed.
- For instructions on Visdom usage/installation, see the Installation section.
- You can pick-up training from a checkpoint by specifying the path as one of the training parameters (again, see
train_refinedet.py
for options)
To evaluate a trained network:
python model_test_one_image.py
You can specify the parameters listed in the model_test_one_image.py
file by flagging them or manually changing them.
We have accumulated the following to-do list, which we hope to complete in the near future
- Still to come:
- Support for multi-scale testing
*[ zhanghang1989 /ResNeSt ] (https://github.com/zhanghang1989/ResNeSt)