UAE's TII Launches Falcon Perception, a Sovereign Computer Vision Model

Abu Dhabi's Technology Innovation Institute (TII), the applied research arm of the Advanced Technology Research Council, announced Falcon Perception in late March 2026 — a next-generation multimodal AI model that performs open-vocabulary visual grounding and segmentation from natural language prompts. The model joins the Falcon family, the UAE's homegrown series of AI models first launched in 2023, and marks a significant step in the country's strategy to develop sovereign AI capabilities across every layer of the technology stack, from language to vision to code.

Despite having just 600 million parameters — significantly fewer than most competing multimodal systems — Falcon Perception scores 68.0 Macro-F1 on the SA-Co benchmark, outperforming Meta's SAM 3 (62.3) with particularly strong gains in food and drink recognition (70.3 vs 58.1), sports (75.2 vs 71.2), and attribute recognition (79.3 vs 71.1). The performance advantage comes from its early-fusion architecture, which unifies image and language processing from the first network layer rather than treating them as separate modalities merged downstream. This design makes the model more efficient per parameter — critical for real-world deployment at scale.

Falcon Perception continues TII's strategy of releasing capable models under permissive open-source licenses to maximize global developer adoption. The global open-source AI model market is growing at over 23% CAGR through 2030, and TII's Falcon series has already established itself as a respected contributor to the international open-source AI ecosystem. By publishing a vision model alongside its existing language, speech, and code models, TII is building out a full-spectrum foundation model library that supports Arabic-language AI applications across modalities.

For the UAE's AI ecosystem, Falcon Perception carries particular strategic weight. Computer vision is a foundational capability for smart city applications, infrastructure monitoring, healthcare diagnostics, and the tourism and retail sectors — all priority areas under the UAE's National AI Strategy 2031. A domestically produced, open-source vision model means government and enterprise deployers can implement computer vision applications with full data sovereignty, without dependency on foreign APIs for inference. This is especially relevant for security-sensitive use cases in government facilities, border management, and critical infrastructure.

Computer vision has been a core service pillar at Diverge since the company's founding. Diverge's computer vision deployments — spanning smart city surveillance, retail analytics, and infrastructure inspection across the UAE and Gulf region — operate in the same domain that Falcon Perception addresses. The emergence of capable, efficient vision models from regional institutions like TII strengthens the wider ecosystem that makes on-premise, sovereignty-compliant computer vision deployments commercially viable. For Diverge's clients, regional foundation models reduce latency, cost, and data governance complexity.

Falcon Perception signals that the UAE is no longer simply a consumer of frontier AI — it is actively contributing to it. As model efficiency continues to improve, compact multimodal models capable of running on-device or at the edge will become the preferred deployment pattern for latency-sensitive computer vision applications. The next phase of UAE AI development will likely see TII expand the Falcon Perception family with domain-specific variants for energy, healthcare, and public safety — further deepening the country's sovereign AI capability stack.