Closed ChuhuaW closed 1 month ago
Hi @ChuhuaW
This is the checkpoint for my final challenge submission.
Your results seem consistent with what I observed during my participation in the challenge. There was a notable discrepancy between the validation and test set results, with the latter generally being higher. Keep in mind that at inference, results can vary depending on yolo_max_det
and top_k_verb
even when using the same checkpoint.
Please let me know if there are any other problems :)
@KeenyJin Thank you so much!
Hey there,
Thanks for sharing your work. I was wondering if you also plan on releasing the checkpoints?
I followed the instructions and got the following results on the ego4d v2 validation set:
Box + Noun mAP: 30.67 Box + Noun + Verb mAP: 14.70 Box + Noun + TTC mAP: 9.21 Box + Noun + Verb + TTC mAP: 4.89
Are these results expected? Thank you.