MVSNet
MVSNet copied to clipboard
Testing with high-resolution images
Hi,
I'm trying to test on the Tanks and Temples dataset. The command is as follows:
export scene=Family
python test.py --dense_folder /path/to/tankandtemples/intermediate/$scene --regularization '3DCNNs' --max_w 1920 --max_h 1056 --max_d 256 --interval_scale 1.0 --pretrained_model_ckpt_path /path/to/tf_model_dtu --ckpt_step 150000
I tested on GPU with 24GB of memory, but the program could not run due to memory consumption. How do I test on the Tanks and Temples dataset using the same conditions as in the paper?
It shouldn't be... I was using a V100 GPU (16GB) on google ml platform to do the experiments.
But also I found tensorflow GPU usage is somewhat unpredictable when using different GPUs...