AutoFlow: Learning a Better Training Set for Optical Flow

2021/04/29 by Deqing Sun, Daniel Vlasic, Sun, Deqing +15 · 4 citations
Computer Science · #Advanced Image Processing Techniques #Advanced Vision and Imaging #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Image Enhancement Techniques

paper · pdf · doi:10.48550/arxiv.2104.14544

openalex publication_date 2021/04/29 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Synthetic datasets play a critical role in pre-training CNN models for optical flow, but they are painstaking to generate and hard to adapt to new applications. To automate the process, we present AutoFlow, a simple and effective method to render training data for optical flow that optimizes the performance of a model on a target dataset. AutoFlow takes a layered approach to render synthetic data, where the motion, shape, and appearance of each layer are controlled by learnable hyperparameters. Experimental results show that AutoFlow achieves state-of-the-art accuracy in pre-training both PWC-Net and RAFT. Our code and data are available at https://autoflow-google.github.io .

Citations

• The Pascal Visual Object Classes (VOC) Challenge
• Gradient-based learning applied to document recognition
• ImageNet Large Scale Visual Recognition Challenge
• ImageNet classification with deep convolutional neural networks
• Performance of optical flow techniques
• Accelerating 3D Deep Learning with PyTorch3D
• RAFT: Recurrent All-Pairs Field Transforms for Optical Flow

Cited by

• Augmented Runtime Collaboration for Self-Organizing Multi-Agent Systems: A Hybrid Bi-Criteria Routing Approach
• Injecting Frame-Event Complementary Fusion into Diffusion for Optical Flow in Challenging Scenes
• FlowSeek: Optical Flow Made Easier with Depth Foundation Models and Motion Bases
• Separating Knowledge and Perception with Procedural Data

Related

• The Devil in the Details: Simple and Effective Optical Flow Synthetic Data Generation
• Models Matter, So Does Training: An Empirical Study of CNNs for Optical Flow Estimation
• Disentangling Architecture and Training for Optical Flow
• FusionNet and AugmentedFlowNet: Selective Proxy Ground Truth for Training on Unlabeled Images
• Temporal Interpolation as an Unsupervised Pretraining Task for Optical Flow Estimation