Skip to content

Latest commit

 

History

History
160 lines (124 loc) · 7.95 KB

README.md

File metadata and controls

160 lines (124 loc) · 7.95 KB
Raw

The P2BNet code is in mmdet/models/detectors/P2BNet.py mmdet/models/roi_heads/P2B_head.py our GUPs: 8 * RTX3090

Prerequisites

install environment following

conda create -n open-mmlab python=3.7 -y
conda activate open-mmlab
# install latest pytorch prebuilt with the default prebuilt CUDA version (usually the latest)
conda install -c pytorch pytorch torchvision -y
# conda install pytorch==1.9.0 torchvision==0.10.0 torchaudio==0.8.0 cudatoolkit=11.1 -c pytorch -c conda-forge
# install the latest mmcv

    ```shell
    pip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/cu111/torch1.9.0/index.html
    ```

# install mmdetection

pip uninstall pycocotools   # sometimes need to source deactivate before, for 
pip install -r requirements/build.txt
pip install -v -e . --user  # or try "python setup.py develop" if get still got pycocotools error
chmod +x tools/dist_train.sh
conda install scikit-image  # or pip install scikit-image

Prepare dataset COCO

  1. download dataset to data/coco
  2. generate point annotation or download point annotation( Baidu Yun passwd:6752 or Google Driver], move annotations/xxx to data/coco/annotations_qc_pt/xxx
  3. or, you can generate annotation file by yourself:

QC Point generation for coco

1.generate QC point annotation

export MU=(0 0)
export S=(0.25 0.25)  # sigma
export SR=0.25 # size_range
export VERSION=1
export CORNER=""
# export T="val"
export T="train"
PYTHONPATH=. python huicv/coarse_utils/noise_data_mask_utils.py "generate_noisept_dataset" \
    "data/coco/annotations/instances_${T}2017.json" \
    "data/coco/coarse_annotations_new/quasi-center-point-${MU[0]}-${MU[1]}-${S[0]}-${S[1]}-${SR}_${VERSION}/${CORNER}/qc_instances_${T}2017_coarse.json" \
    --rand_type 'center_gaussian' --range_gaussian_sigma "(${MU[0]},${MU[1]})" --range_gaussian_sigma "(${S[0]},${S[1]})" \
    --size_range "${SR}"

2.Transfer QC point annotation to 'bbox' and transfer original bbox to 'true_bbox'

the QC point annotation is transfered to 'bbox' with fixed w and h, which is easy for mmdetection reading and dataset pipeline

the original bbox is transfered to 'true_bbox', which is the real box ground-turth

export VERSION=1
export MU=(0 0)
export S=(0.25 0.25)  # sigma
export RS=0.25
export CORNER=""
export WH=(64 64)
export T="train"
PYTHONPATH=. python huicv/coarse_utils/noise_data_utils.py "generate_pseudo_bbox_for_point" \
    "data/coco/coarse_annotations_new/quasi-center-point-${MU[0]}-${MU[1]}-${S[0]}-${S[1]}-${SR}_${VERSION}/${CORNER}/qc_instances_${T}2017_coarse.json"  \
    "data/coco/coarse_annotations_new/quasi-center-point-${MU[0]}-${MU[1]}-${S[0]}-${S[1]}-${SR}_${VERSION}/${CORNER}/qc_instances_${T}2017_coarse_with_gt.json"  \
    --pseudo_w ${WH[0]} --pseudo_h ${WH[1]}

3.For other dataset, we can transform the annotation style to coco json style and use the same way.

Train, Test and Visualization

Take COCO as example

Prepare trained model

  1. move coco dataset (2017 version) or make a soft link to data/coco
  2. download weight from Baidu Yun(passwd:3pfu) or Google Driver , move weights/P2BNet/epoch_12.pth to ../TOV_mmdetection_cache/work_dir/coco/P2BNet/epoch_12.pth move weights/faster/epoch_12.pth to ../TOV_mmdetection_cache/work_dir/coco/Faster/epoch_12.pth

Train

open to the work path: P2BNet/TOV_mmdetection

  1. P2BNet + FasterRCNN

    # [cmd 0] train P2BNet and inference on training set with P2BNet
work_dir='../TOV_mmdetection_cache/work_dir/coco/' && CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 PORT=10000 tools/dist_train.sh configs2/COCO/P2BNet/P2BNet_r50_fpn_1x_coco_ms.py 8 \
--work-dir=${work_dir}  \
--cfg-options evaluation.save_result_file=${work_dir}'_1200_latest_result.json'

# [cmd 1] turn result file to coco annotation fmt
python exp/tools/result2ann.py data/coco/annotations/instances_train2017.json ../TOV_mmdetection_cache/work_dir/coco/_1200_latest_result.json  ../TOV_mmdetection_cache/work_dir/coco/coco_1200_latest_pseudo_ann_1.json

# [cmd 2] train FasterRCNN
	work_dir='../TOV_mmdetection_cache/work_dir/coco/' && CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 PORT=10003 ./tools/dist_train.sh configs2/COCO/detection/faster_rcnn_r50_fpn_1x_coco.py 8 \
--work-dir=${work_dir}'detection/without_weight' --cfg-options data.train.ann_file=${work_dir}'coco_1200_latest_pseudo_ann_1.json'

  2. inference with trained P2BNet to get pseudo box and train FasterRCNN with pseudo box

    # [cmd 0] inference with trained P2BNet to get pseudo box
work_dir='../TOV_mmdetection_cache/work_dir/coco/' && CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 PORT=10000 tools/dist_train.sh configs2/COCO/P2BNet/P2BNet_r50_fpn_1x_coco_ms.py 8 \
--work-dir=${work_dir}  \
--cfg-options  evaluation.save_result_file=${work_dir}'_1200_latest_result.json' load_from=${work_dir}'P2BNet/epoch_12.pth' evaluation.do_first_eval=True runner.max_epochs=0 

# [cmd 1] turn result file to coco annotation fmt
python exp/tools/result2ann.py data/coco/annotations/instances_train2017.json ../TOV_mmdetection_cache/work_dir/coco/_1200_latest_result.json  ../TOV_mmdetection_cache/work_dir/coco/coco_1200_latest_pseudo_ann_1.json

# [cmd 2] train FasterRCNN
	work_dir='../TOV_mmdetection_cache/work_dir/coco/' && CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 PORT=10003 ./tools/dist_train.sh configs2/COCO/detection/faster_rcnn_r50_fpn_1x_coco.py 8 \
--work-dir=${work_dir}'detection/without_weight' --cfg-options data.train.ann_file=${work_dir}'coco_1200_latest_pseudo_ann_1.json'

Test && Visualization

Inference and Visualization of FasterRCNN

  1. run such command for visualization of detection performance on P2BNet-FR

    # move weights/faster/epoch_12.pth to ../TOV_mmdetection_cache/work_dir/coco/Faster/epoch_12.pth
# [cmd 1] 'exp/visulization/image_demo_P2BNet_FR.py' is the entrance of the visualization project. 
##Line 30 and 31 define the model's config file and the checkpoint file path.
    ## config_file = 'configs2/COCO/detection/faster_rcnn_r50_fpn_1x_coco.py'
    ## checkpoint_file = '../TOV_mmdetection_cache/work_dir/coco/Faster/epoch_12.pth'

##Line 34, 35 and 36 define the images to be visualized. Line 37 defines the output folder.
    ## ann_file = 'data/coco/annotations/instances_val2017.json'  # visualize: only val set
    #### in_folder = '{image folder}'  # visualize: the image folder you need or 'data/coco/images/'
    ## img_dir = 'data/coco/images/'
    ## out_dir = '../TOV_mmdetection_cache/work_dirs/center_like/COCO/detection/'

python exp/visulization/image_demo_P2BNet_FR.py

  2. run such command for visualization of P2BNet

     # [cmd 0] inference with trained P2BNet to get pseudo box
 work_dir='../TOV_mmdetection_cache/work_dir/coco/' && CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 PORT=10000 tools/dist_train.sh configs2/COCO/P2BNet/P2BNet_r50_fpn_1x_coco_ms.py 8 \
 --work-dir=${work_dir}  \
 --cfg-options model.roi_head.with_atten=False  model.roi_head.bbox_head.with_loss_pseudo=False evaluation.save_result_file=${work_dir}'_1200_latest_result.json' load_from=${work_dir}'P2BNet/epoch_12.pth' evaluation.do_first_eval=True runner.max_epochs=0 
 
 # [cmd 1] turn result file to coco annotation fmt
 python exp/tools/result2ann.py data/coco/annotations/instances_train2017.json '../TOV_mmdetection_cache/work_dir/coco/_1200_latest_result.json'  '../TOV_mmdetection_cache/work_dir/coco/coco_1200_latest_pseudo_ann_1.json'

 # [cmd 2] visualization of pseudo annotation as P2BNet result
 ##set line 31 as pseudo anotation path: gt_path='../TOV_mmdetection_cache/work_dir/coco/coco_1200_latest_pseudo_ann_1.json'	
 python exp/visulization/vis_p2bnet.py