Shiro-LK
Shiro-LK
hello, are you sure of the coordinate you give into your coco dataset format ? If I remember well, contrary to PASCAL VOC which is xmin ymin xmax and ymax...
@gitathrun Hi, I have exactly the same issue, do you solve your issue ?
@xvjiarui I am also trying to use gc block on classifier such as resnet, vgg16 etc and I would like to be sure that I am doing thing good. First...
@xvjiarui Hi, Thanks for your reply. Do you still have the model trained on imagenet with gcblock ?
Hi thanks for your answer. I will try that and will let you know. Is it possible with this algorithm to do inference after a training ?
any update ? I have the same question
@voznesenskym I am trying to benchmark your approach with torchdynamo but got some error modules. do you know which version torchinductor, torchdynamo and triton are used to make your modification...
you can use the function recognize instead of readtext, but you will need to provide the bounding boxes in a specific format. Otherwise, you can change the thresholds in readtext...
@jqJordan From what I understood : k (9 anchors) are created for each images. The anchors are resized taking in account the downscale of our network. For example, if we...
@cmbowyer13 did you succeed to solve the problem ?