yolov7报错 indices should be either on cpu or on the same device as the indexed tensor (cpu)解决

yolov7报错 indices should be either on cpu or on the same device as the indexed tensor (cpu)解决

问题

报错信息

Traceback (most recent call last):File "/usr/local/code/yolov7_pytroch/yolov7/train.py", line 616, in <module>train(hyp, opt, device, tb_writer)File "/usr/local/code/yolov7_pytroch/yolov7/train.py", line 363, in trainloss, loss_items = compute_loss_ota(pred, (device), imgs)  # loss scaled by batch_sizeFile "/usr/local/code/yolov7_pytroch/yolov7/utils/loss.py", line 585, in __call__bs, as_, gjs, gis, targets, anchors = self.build_targets(p, targets, imgs)File "/usr/local/code/yolov7_pytroch/yolov7/utils/loss.py", line 759, in build_targetsfrom_which_layer = from_which_layer[fg_mask_inboxes]RuntimeError: indices should be either on cpu or on the same device as the indexed tensor (cpu)

分析

通过 搜索资料及文心一言，chatGPT 解答
chatGPT:
The error message you’re seeing, “RuntimeError: indices should be either on CPU or on the same device as the indexed tensor (CPU),” typically occurs in PyTorch or a similar deep learning framework. This error arises when you’re trying to perform an operation that involves tensors on different devices (e.g., CPU and GPU) and the indices used for indexing are not on the same device as the tensor you’re trying to index.

To resolve this issue, you should ensure that the indices you’re using for indexing are on the same device as the tensor you’re trying to index. Here’s how you can do it:

Check Device Compatibility: First, make sure that the tensor you’re trying to index and the indices you’re using are compatible in terms of device. If you’re using a GPU tensor, the indices should also be on the same GPU.

Move Indices to the Correct Device: If your indices are on the wrong device, you can move them to the correct device using the .to(device) method. For example, if you have a GPU tensor gpu_tensor and you want to move the indices to the GPU

文心一言：
这个问题是在运行PyTorch代码时出现的，它说的是索引应该在CPU上，或者在与索引张量相同的设备上（这里是CPU）。这通常是因为你试图在GPU上的张量上使用CPU索引，或者反过来。

解决这个问题的方法是将你的索引移到与张量相同的设备上。例如，如果你的张量在GPU上，你需要确保你的索引也在GPU上。这可以通过使用.to()方法来实现。

解决

修改loss.py代码
修改对应报错的 ComputeLossOTA类下build_targets方法，添加.to(device)
以下三处 添加或者修改对应的代码
添加：device = torch.device(targets.device)

修改：

from_which_layer.s(size=(len(b),)) * i)
修改为
from_which_layer.append((s(size=(len(b),)) * i).to(device))

matching_matrix = s_like(cost)
修改为
matching_matrix = s_like(cost, device=device)

def build_targets(self, p, targets, imgs):#indices, anch = self.find_positive(p, targets)indices, anch = self.find_3_positive(p, targets)#indices, anch = self.find_4_positive(p, targets)#indices, anch = self.find_5_positive(p, targets)#indices, anch = self.find_9_positive(p, targets)device = torch.device(targets.device)matching_bs = [[] for pp in p]matching_as = [[] for pp in p]matching_gjs = [[] for pp in p]matching_gis = [[] for pp in p]matching_targets = [[] for pp in p]matching_anchs = [[] for pp in p]nl = len(p)    for batch_idx in range(p[0].shape[0]):b_idx = targets[:, 0]==batch_idxthis_target = targets[b_idx]if this_target.shape[0] == 0:continuetxywh = this_target[:, 2:6] * imgs[batch_idx].shape[1]txyxy = xywh2xyxy(txywh)pxyxys = []p_cls = []p_obj = []from_which_layer = []all_b = []all_a = []all_gj = []all_gi = []all_anch = []for i, pi in enumerate(p):b, a, gj, gi = indices[i]idx = (b == batch_idx)b, a, gj, gi = b[idx], a[idx], gj[idx], gi[idx]                all_b.append(b)all_a.append(a)all_gj.append(gj)all_gi.append(gi)all_anch.append(anch[i][idx])from_which_layer.append((s(size=(len(b),)) * i).to(device))fg_pred = pi[b, a, gj, gi]                p_obj.append(fg_pred[:, 4:5])p_cls.append(fg_pred[:, 5:])grid = torch.stack([gi, gj], dim=1)pxy = (fg_pred[:, :2].sigmoid() * 2. - 0.5 + grid) * self.stride[i] #/ 8.#pxy = (fg_pred[:, :2].sigmoid() * 3. - 1. + grid) * self.stride[i]pwh = (fg_pred[:, 2:4].sigmoid() * 2) ** 2 * anch[i][idx] * self.stride[i] #/ 8.pxywh = torch.cat([pxy, pwh], dim=-1)pxyxy = xywh2xyxy(pxywh)pxyxys.append(pxyxy)pxyxys = torch.cat(pxyxys, dim=0)if pxyxys.shape[0] == 0:continuep_obj = torch.cat(p_obj, dim=0)p_cls = torch.cat(p_cls, dim=0)from_which_layer = torch.cat(from_which_layer, dim=0)all_b = torch.cat(all_b, dim=0)all_a = torch.cat(all_a, dim=0)all_gj = torch.cat(all_gj, dim=0)all_gi = torch.cat(all_gi, dim=0)all_anch = torch.cat(all_anch, dim=0)pair_wise_iou = box_iou(txyxy, pxyxys)pair_wise_iou_loss = -torch.log(pair_wise_iou + 1e-8)top_k, _ = pk(pair_wise_iou, min(10, pair_wise_iou.shape[1]), dim=1)dynamic_ks = torch.clamp(top_k.sum(1).int(), min=1)gt_cls_per_image = (F.one_hot(this_target[:, 1].to(torch.int64), ).float().unsqueeze(1).repeat(1, pxyxys.shape[0], 1))num_gt = this_target.shape[0]cls_preds_ = (p_cls.float().unsqueeze(0).repeat(num_gt, 1, 1).sigmoid_()* p_obj.unsqueeze(0).repeat(num_gt, 1, 1).sigmoid_())y = cls_preds_.sqrt_()pair_wise_cls_loss = F.binary_cross_entropy_with_logits(torch.log(y/(1-y)) , gt_cls_per_image, reduction="none").sum(-1)del cls_preds_cost = (pair_wise_cls_loss+ 3.0 * pair_wise_iou_loss)matching_matrix = s_like(cost, device=device)for gt_idx in range(num_gt):_, pos_idx = pk(cost[gt_idx], k=dynamic_ks[gt_idx].item(), largest=False)matching_matrix[gt_idx][pos_idx] = 1.0del top_k, dynamic_ksanchor_matching_gt = matching_matrix.sum(0)if (anchor_matching_gt > 1).sum() > 0:_, cost_argmin = torch.min(cost[:, anchor_matching_gt > 1], dim=0)matching_matrix[:, anchor_matching_gt > 1] *= 0.0matching_matrix[cost_argmin, anchor_matching_gt > 1] = 1.0fg_mask_inboxes = matching_matrix.sum(0) > 0.0matched_gt_inds = matching_matrix[:, fg_mask_inboxes].argmax(0)from_which_layer = from_which_layer[fg_mask_inboxes]all_b = all_b[fg_mask_inboxes]all_a = all_a[fg_mask_inboxes]all_gj = all_gj[fg_mask_inboxes]all_gi = all_gi[fg_mask_inboxes]all_anch = all_anch[fg_mask_inboxes]this_target = this_target[matched_gt_inds]for i in range(nl):layer_idx = from_which_layer == imatching_bs[i].append(all_b[layer_idx])matching_as[i].append(all_a[layer_idx])matching_gjs[i].append(all_gj[layer_idx])matching_gis[i].append(all_gi[layer_idx])matching_targets[i].append(this_target[layer_idx])matching_anchs[i].append(all_anch[layer_idx])for i in range(nl):if matching_targets[i] != []:matching_bs[i] = torch.cat(matching_bs[i], dim=0)matching_as[i] = torch.cat(matching_as[i], dim=0)matching_gjs[i] = torch.cat(matching_gjs[i], dim=0)matching_gis[i] = torch.cat(matching_gis[i], dim=0)matching_targets[i] = torch.cat(matching_targets[i], dim=0)matching_anchs[i] = torch.cat(matching_anchs[i], dim=0)else:matching_bs[i] = sor([], device='cuda:0', dtype=torch.int64)matching_as[i] = sor([], device='cuda:0', dtype=torch.int64)matching_gjs[i] = sor([], device='cuda:0', dtype=torch.int64)matching_gis[i] = sor([], device='cuda:0', dtype=torch.int64)matching_targets[i] = sor([], device='cuda:0', dtype=torch.int64)matching_anchs[i] = sor([], device='cuda:0', dtype=torch.int64)return matching_bs, matching_as, matching_gjs, matching_gis, matching_targets, matching_anchs  

其实最新版 github上该代码已修改，原使用代码可能是之前的版本。
.py