Skip to content

Latest commit

 

History

History
101 lines (73 loc) · 2.58 KB

README.md

File metadata and controls

101 lines (73 loc) · 2.58 KB
Raw

AWDesc (Local features detection and description)

Implementation of Attention Weighted Local Descriptors (TPAMI2023).

Unofficial Pytorch implementation of SuperPoint.

To do：

  • Evaluation code and Trained model for AWDesc
  • Training code
  • Training code of SuperPoint
  • More detailed readme (Coming soon)

Requirement

pip install -r requirement.txt,

Quick start

HPatches Image Matching Benchmark

1.Download the trained model:

AWDesc_CA:

https://drive.google.com/file/d/1qrvdd3KVYFl6EwH8s5IS5p_Hs26xIKRD/view?usp=sharing

AWDesc_Tiny:

https://drive.google.com/drive/folders/1PGHiGojkE7qCp1T-l9JSn4aJ7gN0_ua6?usp=sharing

and place it in the "ckpt/mtldesc".

2.Download the HPatches dataset：

cd evaluation_hpatch/hpatches_sequences
bash download.sh

3.Extract local descriptors：

cd evaluation_hpatch
CUDA_VISIBLE_DEVICES=0 python export.py  --tag [Descriptor_suffix_name] --top-k 10000 --output_root [out_dir] --config ../configs/MTLDesc_eva.yaml

4.Evaluation

cd evaluation_hpatch/hpatches_sequences
jupyter-notebook

run HPatches-Sequences-Matching-Benchmark.ipynb

Training

AWDesc-CA

Download dataset: https://drive.google.com/file/d/1Uz0hVFPxWsE71V77kXZ973iY2GuXC20b/view?usp=sharing

Set the dataset path in the configuration file configs/AWDesc_train_CA.yaml

mega_image_dir:  /data/Mega_train/image   #images
mega_keypoint_dir:  /data/Mega_train/keypoint #keypoints
mega_despoint_dir:  /data/Mega_train/despoint #descriptor correspondence points

python train.py --gpus 0 --configs configs/AWDesc_train.yaml --indicator awdesc_ca

SuperPoint

Set the dataset path in the configuration file configs/SuperPoint_train.yaml

mega_image_dir:  /data/Mega_train/image   #images
mega_keypoint_dir:  /data/Mega_train/keypoint #keypoints
mega_despoint_dir:  /data/Mega_train/despoint #descriptor correspondence points

python train.py --gpus 0 --configs configs/SuperPoint_train.yaml --indicator superpoint

AWDesc-Tiny

Download dataset: https://pan.baidu.com/s/1-1rpNxYsNl5fVRKB6EWo4A?pwd=elcb

download code：elcb

Set the dataset path in the configuration file configs/AWDesc_train_Tiny.yaml

mega_image_dir:  /data/Mega_train/image   #images
mega_keypoint_dir:  /data/Mega_train/keypoint #keypoints
mega_despoint_dir:  /data/Mega_train/despoint #descriptor correspondence points
mega_dl_dir1:  /data/Mega_train/dl_teacher0 #Knowledge extracted from teacher
mega_dl_dir2:  /data/Mega_train/dl_teacher1 #Knowledge extracted from teacher

python train.py --gpus 0 --configs configs/AWDesc_train_Tiny.yaml --indicator awdesc_t16