Renan Moreira Pinto

i tryed this and didn't solve my problem always broker on the gxf and i tryed to build it from source [component_container_mt-1] 2024-07-11 17:49:04.094 ERROR gxf/std/program.cpp@551: Attempted interrupting when not...

Got the same problem the train dont look to work . I download the dataset for canned meat that NVSII provide for test and see if i get any changes.....

Start testing with train2 .. i just had to change 2 things : line 140 #parser.add_argument("--local_rank", type=int) to parser.add_argument('--local_rank', '--local-rank', type=int, default=0) and add on line 155 local_rank = opt.local_rank...

To use the nvsii I had to downgrade the nvidia drivers to 450 and use the ubuntu 20.04, but didn't go well at middle of image creation the pc shutdown.....

i got some results using the train2 with 20.000 photos and 60 epochs ...    will increase the dataset and raise the training time.

> Hey @RenanMoreiraPinto > > Are these results with data generated with nvsii and this version of the code https://github.com/NVlabs/Deep_Object_Pose/tree/128631a23c827d2091cfd103c03c8c3a93fc6134 ? > > Could you describe how did you generate...

  [00007.json](https://github.com/user-attachments/files/15992039/00007.json)  [00008.json](https://github.com/user-attachments/files/15992042/00008.json)  [00009.json](https://github.com/user-attachments/files/15992043/00009.json)  [00010.json](https://github.com/user-attachments/files/15992044/00010.json)  i made a run file to run all to make it work i just add...

hey you used ros1? do you have a docker version you using? thanks!