Attribute error in sample_matches.py

[acerdur]

Hello,

Thanks for the great work. Inside matching/feature_based/sample_matches.py the images are loaded by calling get_by_path method of the datasets.

However, the ImageDataset class does not have such an attribute, so this ends up with an error.
Replacing that with the _load_img method and applying ToTensor() on its output solved the issue for me.

For your notice.
Cheers!

[ZouaghiHoussem]

from what i understood, they hardcoded the name GTA in their scripts

[czero69]

Hey, alongside replacing with the _load_img and applying torch.from_numpy(), I also needed to permute channels accordingly, and in the end make cpu detach in matching/feature_based/sample_matches.py. Below is my code:

for x in range(min(25,src_id.shape[0])):
			i = int(rd[x])
			print(f'\tloading sample {i}...')
			#img, _ = src_dataset.get_by_path(src_paths[int(src_id[i])])
			img = torch.from_numpy(src_dataset._load_img(src_paths[int(src_id[i])])).permute(2,0,1)
			print("img size", img.size())
			r0,r1,c0,c1 = src_coords[int(src_id[i])]
			a = img[:,r0:r1,c0:c1].unsqueeze(0)
			#img, _ = dst_dataset.get_by_path(dst_paths[int(dst_id[int(src_id[i]), int(knn[i])])])
			img = torch.from_numpy(dst_dataset._load_img(dst_paths[int(dst_id[int(src_id[i]), int(knn[i])])])).permute(2,0,1)
			r0,r1,c0,c1 = dst_coords[int(dst_id[int(src_id[i]), int(knn[i])])]
			b = img[:,r0:r1,c0:c1].unsqueeze(0)
			crops.append(a)
			crops.append(b)
			pass

		if len(crops) > 0:
			grid = make_grid(torch.cat(crops, 0), nrow=2)
			some_arr = (255.0*grid.permute(1,2,0).cpu().detach().numpy()).astype(np.uint8)
			print("some_arr shp:", some_arr.shape)
			imwrite(f'knn_{t}.jpg', some_arr)
			pass
		pass