TasaNet: temporal attention and spatial auxiliary network for moving Infrared dim and small target detection

Model

TasaNet

PR results on the DAUB and IRDST datasets

DAUB

IRDST

Datasets

1. DUAB and IRDST

2. The COCO format need to convert to txt format.

python utils_coco/coco_to_txt.py

3. The class of dataset should write to a txt file.

Such as model_data/classes.txt

Train

The hyper-parameters are set in train.py

python tarin.py

Evaluate

The hyper-parameters are set in vid_map_coco.py

python vid_map_coco.py

Visualization

The hyper-parameters are setted in vid_predcit.py

python vid_predcit.py

Reference

https://github.com/bubbliiiing/yolox-pytorch/