michelodu

Hello, Thanks for your kind reply. I've just done as much, and while there is progress, there still seems to be a mismatch when I try to run the ZoeDepth...

I hard-coded a few parameters to make it easier to run on the debugger, and I'm further along. No idea why they were not input properly. parser.add_argument("-m", "--model", type=str, default="zoedepth")...

Thanks. I will add it to the environment.yml file. I'm afraid of breaking my Conda build proceeding otherwise.

Thanks for the info. I'll work on it now. Progress. It fails a bit further on. ...mpr/mpr/DepthAnything/metric_depth/train_mono.py", line 173, in mp.spawn(main_worker, nprocs=ngpus_per_node socket.cpp:464] [c10d] The server socket has failed to...

I think that there are some limitations in the current environment.yml file. I noticed that the pytorch 1.13.1 does not overwrite an existing pytorch library on conda, say 2.2.1. if...

I'm going to use this post to track some of the other problems that I encountered so far. I've been getting a socket error as well. This added option seems...

Hi Dung. I've solved that one actually, but I'm using this thread as a document to track what I am seeing, in case anyone in the community wants to re-use...

My last obstacle is making sure that I train on the files of my choosing. This established in DepthAnything/metric_depth/zoedepth/utils/config.py . This config file includes the following code for kitti, which...

I am progressing... I managed to format my datafile as ZoeDepth expects to see it, but CUDA complains about memory... torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 1.03 GiB...

For the community... Mr Dung and I have been interacting on Instragram. He suggested that I play with the batch size in config_zoedepth.json. I tried with values of bs =...