Problems in the learning process of YOLOv5

Problems in the learning process of YOLOv5

Learning blog link: Target detection – teach you how to build your own YOLOv5 target detection platform

Question 1

The following error occurs when installing requirements

ERROR: Could not build wheels for pycocotools, which is required to install pyproject.toml-based projects

Solution: Modify pycocotools>=2.0 to pycocotools-windows>=2.0 in the requirements.txt file

Reason: The maintainer of pycocotools no longer provides support for the Windows platform

Question 2

While running the train.py file to run I get the following error

AssertionError: Image Not Found D:\PycharmProjects\yolov5-hat\VOCdevkit\images\train\000000

Solution: delete the train.cache and val.cache files in the storage label directory

Reason: The original cache file will be used for training by default during training. Since the absolute path of the training set has been changed, it must be deleted to avoid continuing to load the original path

Question 3

The last step of training appears

RuntimeError: result type Float can‘t be cast to the desired output type __int64

Solution: Find the last for function in loss.py that reported an error in version 5.0, and replace it with the last for function in loss.py in yolov5-master version to run normally

        for i in range(self.nl):
            anchors, shape = self.anchors[i], p[i].shape
            gain[2:6] = torch.tensor(shape)[[3, 2, 3, 2]]  # xyxy gain
 
            # Match targets to anchors
            t = targets * gain  # shape(3,n,7)
            if nt:
                # Matches
                r = t[..., 4:6] / anchors[:, None]  # wh ratio
                j = torch.max(r, 1 / r).max(2)[0] < self.hyp['anchor_t']  # compare
                # j = wh_iou(anchors, t[:, 4:6]) > model.hyp['iou_t']  # iou(3,n)=wh_iou(anchors(3,2), gwh(n,2))
                t = t[j]  # filter
 
                # Offsets
                gxy = t[:, 2:4]  # grid xy
                gxi = gain[[2, 3]] - gxy  # inverse
                j, k = ((gxy % 1 < g) & (gxy > 1)).T
                l, m = ((gxi % 1 < g) & (gxi > 1)).T
                j = torch.stack((torch.ones_like(j), j, k, l, m))
                t = t.repeat((5, 1, 1))[j]
                offsets = (torch.zeros_like(gxy)[None] + off[:, None])[j]
            else:
                t = targets[0]
                offsets = 0
 
            # Define
            bc, gxy, gwh, a = t.chunk(4, 1)  # (image, class), grid xy, grid wh, anchors
            a, (b, c) = a.long().view(-1), bc.long().T  # anchors, image, class
            gij = (gxy - offsets).long()
            gi, gj = gij.T  # grid indices
 
            # Append
            indices.append((b, a, gj.clamp_(0, shape[2] - 1), gi.clamp_(0, shape[3] - 1)))  # image, anchor, grid
            tbox.append(torch.cat((gxy - gij, gwh), 1))  # box
            anch.append(anchors[a])  # anchors
            tcls.append(c)  # class

Question 4

After the yolov5 training is completed, the following error occurs when running the detect.py file:

AttributeError: ‘Upsample‘ object has no attribute ‘recompute_scale_factor‘

Solution:

Enter line 154 of the upsampling.py file in the error report, modify the forward function, and set

recompute_scale_factor=self.recompute_scale_factor

just delete

Reason: It may be a pytorch version problem