sam3_video_segmentation

[Amit-Nadam]

In the notebook example: https://github.com/opengeos/segment-geospatial/blob/main/docs/examples/sam3_video_segmentation.ipynb, we saw how we can take a video and extract segmentation and tracking through frames. I would like to know if there is an option to set the video as batching or something else in order not to upload the full video into RAM.

I use an RTX 3090 GPU and thought of using 128 GB of RAM, and Jupyter crashed when I set all the video. I have ~ 41,500 frames in each video (approximately 20 min)

Thank for your time,
Amit

[mohammadrezwankhan]

[Amit-Nadam]

Thank you for your time to write back, it's useful!