Hi, is it intentional that the model is trained in RGB format here [https://github.com/xtcpete/rdd/blob/main/src/dataset/air_ground_dataset.py#L61](https://github.com/xtcpete/rdd/blob/main/src/dataset/air_ground_dataset.py#L61) [https://github.com/xtcpete/rdd/blob/main/src/dataset/megadepth.py#L71](https://github.com/xtcpete/rdd/blob/main/src/dataset/megadepth.py#L71), but during benchmarking, the images are loaded in BGR (cv2 format) [https://github.com/xtcpete/rdd/blob/main/benchmarks/mega_1500.py#L150-L151](https://github.com/xtcpete/rdd/blob/main/benchmarks/mega_1500.py#L150-L151) ?
Hi, is it intentional that the model is trained in RGB format here https://github.com/xtcpete/rdd/blob/main/src/dataset/air_ground_dataset.py#L61 https://github.com/xtcpete/rdd/blob/main/src/dataset/megadepth.py#L71, but during benchmarking, the images are loaded in BGR (cv2 format) https://github.com/xtcpete/rdd/blob/main/benchmarks/mega_1500.py#L150-L151 ?