نسخة أولية وصول مفتوح
DepthEvidence: Unifying Metric Depth Prediction and Geometric Reasoning in Multimodal Language Models
Spatial reasoning with metric constraints requires linking objects to geometric measurements and preserving their numerical content during language reasoning. We present DepthEvidence, a 4B model that uses its own dense metric predictions as object-grounded evidence for language generation. A camera-conditioned decoder …