FreeFlow: A Bias-free Hierarchical Transformer for Optical Flow Estimation

Researchers introduced FreeFlow, a bias-free hierarchical transformer for optical flow estimation that avoids task-specific inductive biases such as correlation volumes, feature warping, and iterative refinement. The work argues that these biases constrain models to predefined heuristics, limiting expressivity and accuracy.

Optical flow methods typically rely on task-specific inductive biases, such as correlation volumes, feature warping, and iterative refinement, among others, to reach high accuracy. While effective, such biases constrain the model to predefined heuristics, which can limit its expressivity and lead to