An event camera reconstruction framework using a DDPM with ControlNet conditioning shows that generic training suits low-variability digits while specialized training suits high-variability faces.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
eBIRD: Event-based Intensity Image Reconstruction Using Controllable Diffusion Models
An event camera reconstruction framework using a DDPM with ControlNet conditioning shows that generic training suits low-variability digits while specialized training suits high-variability faces.