Open felixgeen opened 1 year ago
ModuleNotFoundError: No module named 'torch'
Indicates that you can likely fix this by installing pytorch first
I installed pytorch with conda install -c conda-forge pytorch
and ran the script again with --mask_categories person car
but it errored out somewhere near the end of the process. Is this something to do with the detectron2?
==============================================================================
Global bundle adjustment
==============================================================================
iter cost cost_change |gradient| |step| tr_ratio tr_radius ls_iter iter_time total_time
0 1.709408e+05 0.00e+00 1.79e+06 0.00e+00 0.00e+00 1.00e+04 0 2.92e-02 1.47e-01
CHOLMOD version 3.0.14, Oct 22, 2019: Symbolic Analysis: status: OK
Architecture: Microsoft Windows
sizeof(int): 4
sizeof(SuiteSparse_long): 8
sizeof(void *): 8
sizeof(double): 8
sizeof(Int): 4 (CHOLMOD's basic integer)
sizeof(BLAS_INT): 4 (integer used in the BLAS)
Results from most recent analysis:
Cholesky flop count: 4.4704e+06
Nonzeros in L: 33825
memory blocks in use: 11
memory in use (MB): 0.0
peak memory usage (MB): 0.1
maxrank: update/downdate rank: 8
supernodal control: 1 40 (supernodal if flops/lnz >= 40)
nmethods: number of ordering methods to try: 1
method 0: natural
flop count: 4.4704e+06
nnz(L): 33825
OK
FATAL: command failed
It seems like this error is unrelated to the Detectron2 command since I now get this error if I run --mask_categories
or not. Could his be because I installed pytorch and broke something else in the process?
Global bundle adjustment
==============================================================================
iter cost cost_change |gradient| |step| tr_ratio tr_radius ls_iter iter_time total_time
0 1.569681e+05 0.00e+00 6.75e+12 0.00e+00 0.00e+00 1.00e+04 0 4.98e-02 2.94e-01
CHOLMOD version 3.0.14, Oct 22, 2019: Symbolic Analysis: status: OK
Architecture: Microsoft Windows
sizeof(int): 4
sizeof(SuiteSparse_long): 8
sizeof(void *): 8
sizeof(double): 8
sizeof(Int): 4 (CHOLMOD's basic integer)
sizeof(BLAS_INT): 4 (integer used in the BLAS)
Results from most recent analysis:
Cholesky flop count: 1.1285e+07
Nonzeros in L: 52326
memory blocks in use: 11
memory in use (MB): 0.0
peak memory usage (MB): 0.2
maxrank: update/downdate rank: 8
supernodal control: 1 40 (supernodal if flops/lnz >= 40)
nmethods: number of ordering methods to try: 1
method 0: natural
flop count: 1.1285e+07
nnz(L): 52326
OK
FATAL: command failed
Colmap fails now on every dataset. Is there a potential fix?
Hi, how do you use detectron2 in instant-ngp? What is the command? Thx
I cannot get Detectron2 to install, presumably because I am using Windows? Will there be a future update where there is an alternative tool that works with Windows? Would be very cool!