OpenPCSeg
OpenPCSeg copied to clipboard
why num_points must always bigger than self.num_points
_, inds, inverse_map = sparse_quantize(pc_, return_index=True, return_inverse=True) if self.training and len(inds) > self.num_points: # NOTE: num_points must always bigger than self.num_points raise RuntimeError('droping point') inds = np.random.choice(inds, self.num_points, replace=False)
您好,这部分代码来自于semantickitti_fusion,这里是想说一个sample中的点数必须大于self.num_points,然后进行采样来让每个样本都使用self.num_points个点训练吗,如果是希望采样的话又为什么要返回runtimeerror,如果是希望self.num_points限制数据最大点数,这个 num_points must always bigger than self.num_points又是为什么,这部分有点弄不清楚,希望得到您的解答,感谢。