segment-anything-tensorrt
segment-anything-tensorrt copied to clipboard
Are there any rough performance numbers for the tensorRT version of SAM
Hi, Can folks trying this repo comment on the performance improvements they gained on their specific hardware using this approach?