sparse-ncnet
sparse-ncnet copied to clipboard
problem about test demo
with torch.no_grad(): corr4d, feature_A_2x, feature_B_2x, fs1, fs2, fs3, fs4 = eval_model_reloc( model, {'source_image':src, 'target_image':tgt} )
It has errors that /usr/local/lib/python3.7/dist-packages/MinkowskiEngine/MinkowskiSparseTensor.py in init(self, features, coordinates, tensor_stride, coordinate_map_key, coordinate_manager, quantization_mode, allocator_type, minkowski_algorithm, requires_grad, device) 264 assert ( 265 features.is_cuda == coordinates.is_cuda --> 266 ), "Features and coordinates must have the same backend." 267 268 coordinate_map_key = CoordinateMapKey(
AssertionError: Features and coordinates must have the same backend.