Dear Authors,
Thank you for your excellent work!
While training the Act-Rea sub-network, I downloaded the "lastframe" of the BDD-OIA dataset and encountered an issue. In the "train_25k_images_actions.json" file, the number of action categories is listed as 5, which contrasts with the paper stating 4 action categories. Could you please advise on how to resolve this discrepancy?
Thank you very much for your consideration!
Dear Authors, Thank you for your excellent work! While training the Act-Rea sub-network, I downloaded the "lastframe" of the BDD-OIA dataset and encountered an issue. In the "train_25k_images_actions.json" file, the number of action categories is listed as 5, which contrasts with the paper stating 4 action categories. Could you please advise on how to resolve this discrepancy? Thank you very much for your consideration!