Open anushaGundapaneni opened 6 years ago
same question
I think cpu is ok
It seems like in eval.py, in line 98
"boxes = lanms.merge_quadrangle_n9(boxes.astype('float32'), nms_thres)" must be commented and replaced with line 97 "boxes = nms_locality.nms_locality(boxes.astype(np.float64), nms_thres)" if using tensorflow-cpu
no need to do any changes on eval.py. how to do the training :( Please help me. I am new to tensorflow. When I ran the eval with my images, it detects all the text. But it failed to detect 10% of the text from the image. I would like to do the training with my images.
yes, @neutronest you are right
Whether this code works on cpu or not ? If it works what are the modifications required ?