EN FR
EN FR




Section: Software

MotionLib: A Neural-Fields Model for Motion Estimation

Participants : Pierre Kornprobst [correspondant] , Emilien Tlapale.

MotionLib implements the neural field model of motion estimation described in [52] , using the Python programming language.

Motion integration is the core of the model: It implements a two-layer model with feedbacks that selects and diffuses motion signals. The main structure has been implemented here, allowing contributors to enrich this model easily. Several tools are also provided to visualize and analyze the distributed velocity fields obtained by this approach.