FishHe
Huizhong，你好～ 很激动能看到国人发表如此优秀的文章。 我和我的小伙伴都很期待密集匹配部分的代码～ Hello, Benjamin， The idea of your article is exciting， especially the space-sweeping method with various bands. My friends and I are looking forward to your mapping code....
Thanks for sharing the mapping code. This project is awesome! @zhz218
Same bug i have met. Then, i found i did not run tracking/create_default_local_file.py. python tracking/create_default_local_file.py --workspace_dir . --data_dir ./data --save_dir .
我在copy 代码的时候发现数据转换的代码有问题，不知道是不是相关，放在这里： void tobuffer(const std::vector& imgs, wa::Image* image) { int pos = 0; for (const cv::Mat& img : imgs) { int Lenth = img.rows * img.cols * 3; int nr...
Same problem I have met. One can easily fix the bug by replacing TH_Index_Base with 0. The official solution: https://github.com/pytorch/pytorch/pull/17591.
The code can compile, however can not work on 2080. Hope for a solution.
When I use GTX2080, the program stops at `unsigned int ProgramCU::ReshapeFeatureList(CuTexImage* raw, CuTexImage* out, int* d_featureCount, float keyLocScale)` and i use printf to test `ReshapeFeatureList`: ``` unsigned int res; printf("Test...
Thanks! @stotko It (https://github.com/niessner/BundleFusion/pull/56) really works!
Same question. The program is stuck at the very beginning.
I have tested on Win10 +CUDA 10.2 + VS2013 + GTX 2080 Super. The most important fix is: #56