Closed havanagrawal closed 5 years ago
Thanks Havan, @pshivraj
My thought is that we should clone the mrcnn module and incorporate it into our clomask
repo. That way, we can easily customize any training/inference class if needed. The mrcnn module only has 5 scripts so it shouldn't bloat our repo that much. We can put all our research code in a research
folder, similar to one of their samples/xxx
folder.
When I interned at Clobotics, they copied the giant Caffe repo and developed their own stuff in parallel folders. At the end of the Capstone, I think they would prefer us passing on a custom mrcnn
module with detailed requirements/installation instructions, without referencing back and forth to matterport
repo.
Let's discuss more.
Fixed by #32, #36
Since the matterport implementation is not available via a package manager, we want a way to consistently refer to the library from code.
MASK_CNN_DIR='/Users/havan/Dropbox/CP/Git/MaskRCNN'
We would then add all necessary paths from code, preferably from a central module/lib:import utils import coco ...