convert DoTA Dataset(Traffic Anomaly) into coco-detection like structure.
you may need to create a Soft Link in DoTA/train and DoTA/val through ln -s .., due to coco-style dataset trainers like mmdetection might try to access DoTA/train/Dota/train/<img_name> and DoTA/train/<img_name> at the same time.
Due to download GoogleDisk is to some extent hard for chinese developers, I would offer you a BaiduNetDisk link for DoTA dataset
Or use the link here: https://pan.baidu.com/s/1XF1IIT2eVFiM-Tr72qTHuw?pwd=1919 code: 1919.
I use cat to concatenate seperated folders together, which may caused some error using unzip. I would recommend using 7z on windows/linux to obtain flawless documents. you may find some installation guides here
I am sorry for people use MacOS that they may have to find their way to do similar things, but I am not pretty sure if there are some developers do such things through MacOS. Anyway you are free to pull any requests I guess...