2 papers
cs.CV2026
Mamba2D: A Natively Multi-Dimensional State-Space Model for Vision Tasks
Enis Baty, Alejandro Hernández DÃaz, Rebecca Davidson +2
State-Space Models (SSMs) have emerged as an efficient alternative to transformers, yet existing visual SSMs retain deeply ingrained biases from their origins in natural language p…
cs.CV2025
FlowDet: Unifying Object Detection and Generative Transport Flows
Enis Baty, C. P. Bridges, Simon Hadfield
We present FlowDet, the first formulation of object detection using modern Conditional Flow Matching techniques. This work follows from DiffusionDet, which originally framed detect…