NanoOWL
NanoOWL provides prompt-driven vision inference for edge devices, enabling real-time object detection and tree-structured semantic inference (Tree Prediction). It is suitable for rapid PoC and production deployment in smart manufacturing, retail, public spaces, and industrial monitoring scenarios.
Upstream project: NVIDIA-AI-IOT/nanoowl
- Category: General Edge Vision AI
OWL-ViT
- OWL-ViT (Open-Vocabulary Detection)
Uses natural language descriptions for target classes, enabling open-vocabulary object detection without retraining. - Tree Prompt Inference Pipeline
Supports nested prompts so a single pipeline can perform object detection and fine-grained attribute/sub-object recognition. - TensorRT Optimization
Uses TensorRT engine acceleration to improve inference efficiency while maintaining real-time performance and deployment feasibility.
Supported Platform
| Platform | Hardware Spec | OS | Edge AI SDK |
|---|---|---|---|
| AIR-075 | NVIDIA Jetson Thor - RAM: 128/64 GB, Storage: 512 GB | JetPack 7.1 | Install |