Running DEMO on multiple GPUs

[maffan-96]

Hello, Is it possible to run demo.py on multiple GPUs (say 2). When I try parallel=True for VisualizationDemoCustom class, it fails to run raising error: Traceback (most recent call last): File "/home/affanm/miniconda3/envs/oneformer_py38_ros/lib/python3.8/site-packages/rospy/topics.py", line 750, in _invoke_callback cb(msg) File "/home/affanm/sensor fusion xtreme/catkin_ws/src/oneformer_pkg/src/demo_ros.py", line 114, in image_callback predictions, vis_output = self.demo.run_on_image(cv_image, self.task) File "/home/affanm/sensor fusion xtreme/catkin_ws/src/oneformer_pkg/scripts/util/local_predictor.py", line 59, in run_on_image predictions = self.predictor(image, task) TypeError: call() takes 2 positional arguments but 3 were given

I tried to look into the documentation, but no information is provided about multiple gpu usage for evaluation with pretrained model. Thanks

[maffan-96]

Commenting to keep threat open

[praeclarumjj3]

Hi @maffan-96, thanks for your interest and patience. Unfortunately, the current demo.py does not support multi-gpu visualization.