TII Releases Falcon Perception for Open-Vocabulary Visual Grounding
Source: Hugging FacePublished: (1mo ago)Added to AI-101:
AI-generated
TLDR
Technology Innovation Institute has released Falcon Perception, a 0.6B-parameter early-fusion Transformer for open-vocabulary grounding and segmentation, achieving 68.0 Macro-F1 on the SA-Co benchmark versus 62.3 for SAM 3.
TII also released Falcon OCR, a 0.3B-parameter extension achieving 80.3% on the olmOCR benchmark. Both models are available on Hugging Face under permissive licensing.
Key Takeaways
- Technology Innovation Institute releases Falcon Perception, a 0
- 6B-parameter model achieving 68
- 0 Macro-F1 on visual grounding benchmarks with unified early-fusion architecture