LongE2V: Long-Horizon Event-Based Video Reconstruction, Prediction, and Interpolation with Diffusion Priors

Forum topic · 小凯 · 2026-07-12

Summary

LongE2V is a new framework that recovers high-quality video from sparse event camera streams by leveraging pre-trained video diffusion priors. It jointly handles three tasks: event-based video reconstruction, prediction, and frame interpolation. By fine-tuning a foundational video model, LongE2V achieves high data efficiency and superior perceptual quality compared to regression methods, which tend to blur textures, and existing generative models that struggle with long-term stability. The authors introduce Autoregressive Unrolling and Adaptive Context Switching to mitigate temporal drift over extremely long sequences, plus Reencoding Alignment with Cross Residual Correction for precise bidirectional consistency in frame interpolation. Event Voxel Density Augmentation provides robustness across different sensor resolutions. Experiments on real-world benchmarks show state-of-the-art performance on all three tasks with strong temporal consistency and zero-shot generalization.

Paper Overview

Research Area: Computer Vision (CV) Authors: Cheng-De Fan, Chun-Wei Tuan Mu, Chen-Wei Chang, Chin-Yang Lin, Kun-Ru Wu, Yu-Chee Tseng, Yu-Lun Liu Published: 2026-07-09 arXiv: 2607.08770

Summary

Recovering high-quality video from sparse event streams is a challenging task. Regression methods often blur textures, while existing generative models struggle with long-term stability. The authors propose LongE2V, a novel approach that leverages pre-trained video diffusion priors to jointly handle event-based video reconstruction, prediction, and frame interpolation.

Key Contributions

Diffusion-prior fine-tuning: By fine-tuning a foundational video model, LongE2V achieves high data efficiency and superior perceptual quality.
Autoregressive Unrolling and Adaptive Context Switching: These techniques mitigate temporal drift in extremely long sequences, enabling long-horizon generation.
Reencoding Alignment with Cross Residual Correction: Ensures precise bidirectional consistency during frame interpolation.
Event Voxel Density Augmentation: Ensures robustness across different event sensor resolutions.

Results

Extensive experiments on real-world benchmarks demonstrate that LongE2V outperforms state-of-the-art methods on all three tasks — reconstruction, prediction, and interpolation — showing excellent temporal consistency and zero-shot generalization.

LongE2V: Long-Horizon Event-Based Video Reconstruction, Prediction, and Interpolation with Diffusion Priors

Summary

Paper Overview

Summary

Key Contributions

Results

Links

Tags