Open Hrren opened 6 months ago
The naming format for LasHeR was quite disorganized. We have revised the naming format but forgot to note that. We sincerely apologize for this oversight. The reference format is as follows:
According to your format, I still encountered the following problem [ WARN:0@4.883] global loadsave.cpp:241 findDecoder imread_('/home/robot/download/LasHeR0327/TrainingSet/trainingset/left11/visible/000165.jpg'): can't open/read file: check file path/integrity
Reffe to TBSI, change the fuction "_get_frame_path" as following for reading image in BAT-main/lib/train/dataset/lasher.py:
def _get_frame_path(self, seq_path, frame_id):
vis_frame_names = sorted(os.listdir(os.path.join(seq_path, 'visible')))
inf_frame_names = sorted(os.listdir(os.path.join(seq_path, 'infrared')))
return os.path.join(seq_path, 'visible', vis_frame_names[frame_id]), os.path.join(seq_path, 'infrared', inf_frame_names[frame_id])
Excellent work! Some images in the Lasher dataset, such as v001.jpg, may become '000001. jpg' when read, resulting in incorrect loading. How can I resolve this issue?