ucbdrive / few-shot-object-detection

Implementations of few-shot object detection benchmarks
Apache License 2.0
1.08k stars 225 forks source link

VOC datasets: train all instances of novel classes #116

Closed cats0212 closed 2 years ago

cats0212 commented 3 years ago

I want to train all the instances of novel classes. For example airplane have 916 instances, can I directly modify _box_1shot_aeroplanetrain.txt to write 916 image paths to it? Then run

python3-m tools.train_net --num-gpus 8 \ --config-file configs/PascalVOC-detection/split1/faster_rcnn_R_101_FPN_ft_novel1_1shot.yaml \ Opts model.weights checkpoints/ VOC/FASTER_RCNN/FASTER_RCNN_R_101_FPN_ALL1 /model_reset_remove.pth

Is the training one instance of airplane or 916 instances of airplane?

thanks.

thomasehuang commented 2 years ago

Apologies for the late response. I believe that should work, as the code will load in all the images in the specified .txt file. You can try it and let me know if that works.

Will close for now. Please feel free to reopen if it is still an issue.