TuSimple
mask-rcnn how to allocate memory, If my own dataset has some big size images like 7127*7127, it has cudo memory error.
mask-rcnn how to allocate memory, If my own dataset has some big size images like 7127*7127, it has cudo memory error. Mask-rcnn allocate memory for computing images in terms of max_shape and infer_shape? How to use them to allocate memory?
It seems that your input image is too large. Try using a small image to test.
@huangzehao , I want to use my own dataset to train the mask rcnn. Actually, the size of images in the dataset varies from 500*1000 to 7000 * 7000. Do you think it is a good idea to resize all the images to the same size? I do not know how to cope with the problem. Because some sizes of images are very big. In the code, it has inferred the shape and maximum shape of images. Do you have some advice? thanks