Open paucarre opened 3 months ago
Unfortunately, there is currently no built-in function to perform inference on a given image using a set of point coordinates directly. However, you could potentially implement this feature by referring to the interact_test
function starting from Line 335 in focsam/segmentors/mixseg_refine.py
. Since the entire codebase utilizes the OpenMMLab libraries, it requires a good understanding of these libraries. We will consider adding such functionality in the future.
Is there anywhere in the code a way to do inference on a given image given a set of point coordingates on the image?