tutor for CNV-BNN_Road-Signs? about bnn-pynq

[giuliogamba]

Hi, you should pick the set of s values (in the example 64 and 96) that both works for your input image size and the max and min size of the object you want to recognize in the scene.

[goldinoragung]

and i tried to re-run the road signs with same picture and there's no box in it. is it possible that actually error?

[giuliogamba]

It is possible, the neural network does not have an 100% accuracy. It is important nonetheless that the tiling of the original image is performed correctly. This means making sure that the crop of the original image is done such that one or more tiles completely embed the object you want to recognize.

[goldinoragung]

Hi Mr. giuliogamba,
Now i know what should i do, i appreciate with your answer and i'm very happy to discuss it again with you.
Thank You So Much