This paper presents a new database and evaluation methodology for optical flow algorithms. The authors propose four types of data to test different aspects of optical flow algorithms: (1) sequences with nonrigid motion where the ground-truth flow is determined by tracking hidden fluorescent texture, (2) realistic synthetic sequences, (3) high frame-rate video used to study interpolation error, and (4) modified stereo sequences of static scenes. In addition to the average angular error used by Barron et al., they compute the absolute flow endpoint error, measures for frame interpolation error, improved statistics, and results at motion discontinuities and in textureless regions. The data is freely available on the web at http://vision.middlebury.edu/flow/. The authors also extend the set of performance measures and the evaluation methodology of Barron et al. to focus attention on current algorithmic problems. They describe the design and collection of their database, the evaluation metrics, and the experimental results. The paper also discusses related work and the taxonomy of optical flow algorithms, including the data term, prior term, and optimization algorithms. The authors conclude that their database and evaluation methodology provide a comprehensive way to evaluate current optical flow algorithms and highlight the challenges associated with complex natural scenes.This paper presents a new database and evaluation methodology for optical flow algorithms. The authors propose four types of data to test different aspects of optical flow algorithms: (1) sequences with nonrigid motion where the ground-truth flow is determined by tracking hidden fluorescent texture, (2) realistic synthetic sequences, (3) high frame-rate video used to study interpolation error, and (4) modified stereo sequences of static scenes. In addition to the average angular error used by Barron et al., they compute the absolute flow endpoint error, measures for frame interpolation error, improved statistics, and results at motion discontinuities and in textureless regions. The data is freely available on the web at http://vision.middlebury.edu/flow/. The authors also extend the set of performance measures and the evaluation methodology of Barron et al. to focus attention on current algorithmic problems. They describe the design and collection of their database, the evaluation metrics, and the experimental results. The paper also discusses related work and the taxonomy of optical flow algorithms, including the data term, prior term, and optimization algorithms. The authors conclude that their database and evaluation methodology provide a comprehensive way to evaluate current optical flow algorithms and highlight the challenges associated with complex natural scenes.