batch inference issue

[stigma0617]

Hi,

I tried to infernce batch images using these lines in refinedet_demo.py

for idx, img in enumerate(im_names):
    ...
    net.blobs['data'].data[idx, :, :,:] = transformed_image

detections = net.forward()['detection_out']

for idx in range(len(im_names)):
    det_label = detection[idx, 0, :, 1] # error occurs!!

This code results in errors because the detection dimension is [1,1,#outputs, 7]

I expected the first dimension is the number of batch and tried to split each detection result per image.

What should I do to inferecne batch images?

[sfzhang15]

@stigma0617 The number of batch is not the first dimension. It is the first index of the final dimension as this.

[stigma0617]

@sfzhang15 thanks for your help!