AIBOX-OrinNano is equipped with NVIDIA's original Jetson Orin Nano core module (8GB RAM version), providing up to 67 TOPS of computing power, supporting private deployment of mainstream AI models such as Transformer and ROS robot models, thereby realizing larger and more complex deep neural networks, and realizing visual development tasks such as object recognition, target detection and tracking, and speech recognition, meeting higher-level AI application scenario requirements.

• Product Datasheet: Click Here
• Product Wiki: Click Here

Application Cases 

• Background Removal via U²-Net
• Monocular Depth Estimation

Feature

• High-performance Edge Computing Module: The NVIDIA Jetson Orin Nano edge computing module (8GB version) is equipped with a six-core ARM CPU and a 1024-core NVIDIA Ampere architecture GPU (including 32 Tensor Cores), which can run multiple AI application pipelines simultaneously and provide powerful inference performance.
  
• Multiple Deep Learning Frameworks: It supports a variety of cuDNN-accelerated deep learning frameworks, including PaddlePaddle, PyTorch, TensorFlow, MATLAB, MxNet, Caffe2, Chainer, Keras, etc., as well as custom operator development, and supports Docker containerization technology.
  
• Aluminum Alloy Enclosure With Efficient Heat Dissipation: AIBOX adopts an industrial-grade all-metal shell and an aluminum alloy structure with good thermal conductivity. The grille design on the side of the top cover is conducive to external airflow and efficient heat dissipation, ensuring computing performance and stability even in high-temperature operating environments.

Application

• AI Software Stack And Ecosystem: NVIDIA JetPack, Isaac ROS, and reference AI workflows seamlessly integrate cutting-edge technologies into your products without expensive in-house AI resources. Experience end-to-end acceleration of AI applications and get to market faster using the same powerful technology that drives data center and cloud deployments.
• The Private Deployment Of Large Models: Large language model: supports the Ollamalocal large model deployment framework and private deployment of ultra-large-scale parameterized models: Llama3, Phi-3 Mini. Vision model: supports EfficientVIT, NanoOWL, NanoSAM, SAM, TAM. AI painting: supports ComfyUIgraphical deployment framework and private deployment of Flux, and a stable Diffusion image generation model in the AIGC field.
  

Hardware Overview