Firstly, I would like to thank you for your sharing. I have carefully reviewed your code, and the training process is consistent with the paper. However, I am not quite clear on how the "Unknown" class is identified during the inference process. It seems that only FastRCNNOutputLayers have been used?
Additionally, I have loaded the pre-trained file you shared and executed "sh test_ood.sh". But the results do not seem to be very good? I am not quite sure about the specific reasons.
Is pretesting necessary, and is there a need to modify the configuration file after pretesting?
Firstly, I would like to thank you for your sharing. I have carefully reviewed your code, and the training process is consistent with the paper. However, I am not quite clear on how the "Unknown" class is identified during the inference process. It seems that only FastRCNNOutputLayers have been used? Additionally, I have loaded the pre-trained file you shared and executed "sh test_ood.sh". But the results do not seem to be very good? I am not quite sure about the specific reasons. Is pretesting necessary, and is there a need to modify the configuration file after pretesting?