I have trained a network for detection using bounding boxes. During evaluation I remove boxes using IoU technique and thresholding. However, I am getting multiple boxes for a single object (as seen in picture below). The bounding boxes are
tensor([[448.00, 219.25, 476.33, 287.26], [453.84, 223.89, 473.85, 279.77], [450.89, 218.27, 467.31, 276.45], [448.76, 219.92, 484.23, 283.18]])
Is there any way i can combine or remove the extra boxes?
