UAVDB-TwoFrame: An Extension of UAVDB with Consecutive Frame Pairs

This dataset is an extension of the UAVDB, organized as consecutive frame pairs to support tasks involving temporal information. It includes: 10,763 * 2 images in train/ 2,720 * 2 images in val/ 4,578 * 2 images in test/ Each frame pair is stored using the naming pattern: d{a}c{b}_{n}.jpg, d{a}c{b}_{n+1}.jpg For example, files in the train/ folder include: d1c0_0.jpg, d1c0_1.jpgd1c0_10.jpg, d1c0_11.jpg...d4c5_25211.jpg, d4c5_25212.jpg This format enables optical flow estimation and temporal deep learning applications, offering enhanced flexibility for training, evaluation, or reproduction of motion-aware vision tasks.