MEMFOF: High-Resolution Training for Memory-Efficient Multi-Frame Optical Flow Estimation

TL;DR

MEMFOF introduces a memory-efficient multi-frame optical flow method that achieves high accuracy at high resolutions with significantly reduced GPU memory requirements, enabling training and inference on full HD images.

Contribution

The paper presents MEMFOF, a novel high-resolution training approach for multi-frame optical flow that reduces memory usage while maintaining state-of-the-art accuracy.

Findings

Requires only 2.09 GB GPU memory for 1080p inference

Achieves top performance on multiple benchmarks

Outperforms resource-intensive methods in accuracy and efficiency

Abstract

Recent advances in optical flow estimation have prioritized accuracy at the cost of growing GPU memory consumption, particularly for high-resolution (FullHD) inputs. We introduce MEMFOF, a memory-efficient multi-frame optical flow method that identifies a favorable trade-off between multi-frame estimation and GPU memory usage. Notably, MEMFOF requires only 2.09 GB of GPU memory at runtime for 1080p inputs, and 28.5 GB during training, which uniquely positions our method to be trained at native 1080p without the need for cropping or downsampling. We systematically revisit design choices from RAFT-like architectures, integrating reduced correlation volumes and high-resolution training protocols alongside multi-frame estimation, to achieve state-of-the-art performance across multiple benchmarks while substantially reducing memory overhead. Our method outperforms more resource-intensive alternatives in both accuracy and runtime efficiency, validating its robustness for flow estimation at high resolutions. At the time of submission, our method ranks first on the Spring benchmark with a 1-pixel (1px) outlier rate of 3.289, leads Sintel (clean) with an endpoint error (EPE) of 0.963, and achieves the best Fl-all error on KITTI-2015 at 2.94%. The code is available at https://github.com/msu-video-group/memfof.