# In order to facilitate script management, I have integrated the code into YOLOX-OBB #
Firstly, create python environment
$ conda create -n yolox_inst python=3.7 -y
Then, clone the github of the item
$ git clone https://github.com/DDGRCF/YOLOX-CondInst.git
Then, you can adjust follow the original quick start
I prepare the shell the demo script so that you can quickly run obb demo as:
$ cd my_exps
$ bash demo_inst.sh 0 /path/to/you
# PS: 0 is to assign the train environment to 0 gpu, you can change it by youself and /path/to/you is your demo images.
I define the model default training parameters as following:
model | max epoch | enable_mixup | enable_mosaic | no aug epoch |
---|---|---|---|---|
yolox_s | 24 | True | True | 5 |
cls_loss_weight | obj_loss_weight | iou_loss_weight | reg_loss_weight | mask_loss_weight |
---|---|---|---|---|
1.0 | 1.0 | 5.0 | 1.0 | 5.0 |
Of course, this group parameters is not the best one, so you can try youself. And for the quick train, I have prepare the shell scripts, too.
$ cd my_exps
$ bash train_dota_obb.sh 0
As I set parameters above with 16 batch size per gpu (2gpu), the lresults on val dataset show as following: waiting ...
I just follow original evaluation to test and eval
$ cd my_exps
$ ./eval_dota_obb.sh eval/test 0
# PS: for convenience, I set default parameters. So, eval means evaluating COCO val datasets.