Inference time

[Kyeonga-Kim]

Hi, I've been following your code and paper with interest! Just to let you know, I checked the inference time with your test_wider.py code with widerface db (after resizing to vga size) and it doesn't 37.7 ms. Can you tell me how you measured it?

Below is an example of where I took the measurement points. Thanks :)

        t1 = time.time()

        # NOTE: modified
        # det0 shape: 750, 5
        det0 = detect_face(net, img, shrink, thresh=thresh)

        # multi-scale
        if not single_scale_test:
            print('det1')
            det1 = flip_test(net, img, shrink, thresh=thresh)  # flip test

            print('det2')
            [det2, det3] = multi_scale_test(net, img, max_im_shrink, thresh=thresh)

            det = np.row_stack((det0, det1, det2, det3))
            dets = bbox_vote(det)
        else:
         
            dets = det0

        t2 = time.time()

        print("Detect %04d th image costs %.4f sec" % (counter, t2 - t1))