RuntimeError: random_ expects 'from' to be less than 'to', but got from=0 ＞= to=0

问题：ES6D项目，运行`python train.py —loss_type GADD`，出现报错：

./datasets/tless/models/.pcd
[Open3D WARNING] Read PCD failed: unable to open file: ./datasets/tless/models/.pcd
tensor([], size=(1, 0, 3))
Traceback (most recent call last):
  File "train.py", line 475, in <module>
    main()
  File "train.py", line 116, in main
    per_processor(0, opt)
  File "train.py", line 232, in per_processor
    dataset = pose_dataset('train', opt.num_points, opt.dataset_root, True, opt.noise_trans)
  File "/root/autodl-tmp/ES6D/datasets/tless/tless_dataset.py", line 64, in __init__
    xyz_ids = farthest_point_sample(raw_xyz, num_pt).cpu().numpy()
  File "/root/autodl-tmp/ES6D/datasets/tless/tless_dataset.py", line 222, in farthest_point_sample
    farthest = torch.randint(0, N, (B,), dtype=torch.long).to(device)
RuntimeError: random_ expects 'from' to be less than 'to', but got from=0 >= to=0

解决方法：

在”/root/autodl-tmp/ES6D/datasets/tless/tless_dataset.py”文件中，在`while 1`里添加一个检查项：

        while 1:
            class_input = class_file.readline()
            if not class_input:
                break
            input_cloud = o3d.io.read_point_cloud('{0}/models/{1}.pcd'.format(self.root, class_input[:-1]))
            raw_xyz = torch.tensor(np.asarray(input_cloud.points).reshape((1, -1, 3)), dtype=torch.float32)
            # 添加if not 检查
            if not np.asarray(input_cloud.points).any():
                print("No points in point cloud for class: ", class_input)
                continue
            xyz_ids = farthest_point_sample(raw_xyz, num_pt).cpu().numpy()
            raw_xyz = np.asarray(input_cloud.points).astype(np.float32) * 0.001
            self.cld.append(raw_xyz[xyz_ids[0, :], :])

---

来源：CSDN 原文，首次发布于 2024-05-24。