dshahrokhian / YOLO_tensorflow

:oncoming_taxi: Object Detection (YOLOv1) implentation in tensorflow, with training, testing and video features.
42 stars 22 forks source link

I don't understand this code. Please help me! #9

Closed vietnvri closed 4 years ago

vietnvri commented 5 years ago

x[i][j][index]= (float(prelabel[i][j][index][0])/len(img))*7-i y[i][j][index] = (float(prelabel[i][ j][ index][ 1])/len(img[0]))*7-j w[i][j][index] = np.sqrt(prelabel[i][ j][ index][ 2])/len(img)*7 h[i][j][index] = np.sqrt(prelabel[i][ j][ index][ 3])/len(img[0])

Why x and y are subtract with -i, -j and w is multiply 7? And why activation output fc_32 is linear while x in range [0, 1]?

Thanks for your explanation! Vietnv