VanChilly comments

Results 5 comments of


                                            VanChilly

Acc1 about training HMDB51 pretrained on Imagenet

this is my command ```shell python -u main.py hmdb51 RGB --arch resnet50 --num_segments 8 --gd 20 --lr 0.001 --lr_steps 10 20 --epochs 25 --batch-size 8 -j 8 --dropout 0.8 --consensus_type=avg...

How to get the bounding box cor ordinates of the original image size ?The image is resized and bounding box is applied on it ,but in final output the original image and bounding box of resized image

> In demo.py, the resize_image() function return **re_im, (rh, rw).** > when you get the box corordinates, it is like this:[x1, y1, x2, y2, x3, y3, x4, y4, scores], these...

Acc1 about training HMDB51 pretrained on Imagenet

> same here, maximum 51% acc for 50 epochs at [20, 40] decay. So, it's clear that the result is reasonable for HMDB pre-trained on ImageNet. 👍

Acc1 about training HMDB51 pretrained on Imagenet

> @VanChilly I think the result are obtained by a pre-trained model from K400 then finetune on HMDB Yeah, the author mentioned that in the paper. Thank you.

Could you please share the something v1 dataset?

Yes, I meet the same problem, could anyone share me a link to download this datasets, thx.