segment-anything
segment-anything copied to clipboard
Grayscale image segmentation
Great job, can I input a grayscale map into the SAM model instead of a three-channel color image?
The following works for now:
image = np.expand_dims(image, axis=2)
image = np.repeat(image, 3, axis=2)
With this, I assume there is no benefit in performance to running grayscale?
我猜@dvdrplus的意思是说加一个灰度图的模块
I guess @ dvdrplus means adding a grayscale module 似乎SAM的数据集里面没包括医学影像 It seems that the SAM dataset does not include medical imaging 但是,我试了一下,对于医学影像SAM做的也不错👍🏻 But I gave it a try and did a good job with medical imaging SAM 👍🏻 所以, 有必要做个这样的模块专门处理像xRay, CT, MRI等这样的灰度图 So, it is necessary to create a module specifically for processing grayscale images such as xRay, CT, MRI, etc