Fast variant of TopicFM with 2-head attention and full attention mechanism.
from matching import get_matcher matcher = get_matcher("topicfm", device="cuda") mkpts0, mkpts1, _, _, _, _ = matcher(img0, img1)
