KPConv-STPLS3D- Data Problem

[zhangbaobao1027]

first of all thanks. I use KPConv and STPLS3D dataset for training， When the training progresses to epoch130, an error will appear: 'It seems this dataset only containes empty input spheres'

Currently I skip these abnormal epochs, is it a data problem, and is there a better solution? part of the code：

        # Number collected
        n = input_inds.shape[0]

        # Safe check for empty spheres  
        if n < 2:
            failed_attempts += 1
            if failed_attempts > 100 * self.config.batch_num:
                # raise ValueError('It seems this dataset only containes empty input spheres')
                print('It seems this dataset only containes empty input spheres')
                return []
            t += [time.time()]
            t += [time.time()]
            continue

        # Collect labels and colors
        input_points = (points[input_inds] - center_point).astype(np.float32)

[meidachen]

@zhangbaobao1027 , Really sorry for the late reply, It seems like the point density and the sphere radius make it empty sometimes. I believe this should work fine, but just to make sure the density of the points and the radius are making sense.

[zhangbaobao1027]

Thank you very much, it's real a data problem.