Bias for Action: Video Implicit Neural Representations with Bias Modulation

TL;DR

ActINR introduces a novel implicit neural representation framework for videos, leveraging bias modulation to enable high-quality video synthesis, super-resolution, and inpainting with significant improvements over existing methods.

Contribution

The paper presents a new continuous video modeling method using bias modulation in INRs, allowing shared weights and learned biases for improved video processing tasks.

Findings

Achieves 10x slow motion and 4x spatial super-resolution.

Improves video quality by over 6dB in many tasks.

Sets new standards in continuous video modeling.

Abstract

We propose a new continuous video modeling framework based on implicit neural representations (INRs) called ActINR. At the core of our approach is the observation that INRs can be considered as a learnable dictionary, with the shapes of the basis functions governed by the weights of the INR, and their locations governed by the biases. Given compact non-linear activation functions, we hypothesize that an INR's biases are suitable to capture motion across images, and facilitate compact representations for video sequences. Using these observations, we design ActINR to share INR weights across frames of a video sequence, while using unique biases for each frame. We further model the biases as the output of a separate INR conditioned on time index to promote smoothness. By training the video INR and this bias INR together, we demonstrate unique capabilities, including $10\times$ video slow motion, 4x spatial super resolution along with 2x slow motion, denoising, and video inpainting. ActINR performs remarkably well across numerous video processing tasks (often achieving more than 6dB improvement), setting a new standard for continuous modeling of videos.