Can we use 1280x720 images to train a segmentation model for my custom dataset?

gaomingqi / Track-Anything

Track-Anything is a flexible and interactive tool for video object tracking and segmentation, based on Segment Anything, XMem, and E2FGVI.

MIT License

6.51k stars 481 forks source link

Can we use 1280x720 images to train a segmentation model for my custom dataset? #128

Open Muthukumar4796 opened 9 months ago

Muthukumar4796 commented 9 months ago

Is it feasible to train a segmentation model for a custom dataset using 1280x720 images? I'm working on a project that requires real-time object detection and segmentation from a live camera feed.