Open zuguang0425 opened 10 months ago
Hi, you inserted the GMAC evaluation code in the training loop. During training, the length of the temporal feature is not fixed and might be shorter than 2304. You can try inserting this part in the test part on THUMOS.
我们在train utils.py 中 valid_one_epoch函数中添加了测试GMACs代码,得到如下结果:
这样计算是否正确
This result of MAC is too small. Can you check the input shape and the model setting, and try to evaluate each module inside the forward() function in meta_arch.py?
我们在train utils.py 中 train_one_epoch函数中添加了测试GMACs代码,得到如下结果:
请问论文中表6是如何计算得到的,希望得到你的帮助,非常感谢