Now, I have to find some objects at a matrix, the shape is (x,y) rather than image shape (x,y,3), so the pre-trained model isn't suitable for my model.
And I try to convert the matrix into image, but the result is bad (maybe I do something wrong, so I will try again). So how can I train faster rcnn without pre-trained model?
Now, I have to find some objects at a matrix, the shape is (x,y) rather than image shape (x,y,3), so the pre-trained model isn't suitable for my model. And I try to convert the matrix into image, but the result is bad (maybe I do something wrong, so I will try again). So how can I train faster rcnn without pre-trained model?