taozhuo2000
commented
9 months ago +-------------+-------------+-------------+-------------+-------------+-------------+ | R@1,IoU@0.3 | R@1,IoU@0.5 | R@1,IoU@0.7 | R@5,IoU@0.3 | R@5,IoU@0.5 | R@5,IoU@0.7 | +-------------+-------------+-------------+-------------+-------------+-------------+ | 57.21 | 34.16 | 16.99 | 88.04 | 70.99 | 50.41 | +-------------+-------------+-------------+-------------+-------------+-------------+ that is my result. It also have lower results.
Your work is outstanding and has been highly inspiring to me. However, I am encountering a deviation of approximately 15% lower results compared to the metrics provided in the paper on the ActivityNet dataset. I'm wondering if the hyperparameters in the configuration file are correct, or if I might have overlooked some other crucial factors that require attention? +-------------+-------------+-------------+-------------+-------------+-------------+ | R@1,IoU@0.3 | R@1,IoU@0.5 | R@1,IoU@0.7 | R@5,IoU@0.3 | R@5,IoU@0.5 | R@5,IoU@0.7 | +-------------+-------------+-------------+-------------+-------------+-------------+ | 45.32 | 27.53 | 15.42 | 71.59 | 42.66 | 27.70 | +-------------+-------------+-------------+-------------+-------------+-------------+