DEEPX AI
The Digital View ALC-4096-AIH and AI-100 boards support the DeepX DX-M1 accelerator module. This provides 25 TOPs and is designed for exceptional power efficiency.
We have tested the DeepX DX-M1 module with 4 video streams, and it works well. We will be scheduling more tests in the near future. See resource links below:
The likely applications for the DeepX AI accelerator include:
- High-resolution video analysis
- Edge servers
- Robotics
- Industrial computing
A few more technical details:
- Supported AI frameworks: Tensorflow, Pytorch, ONNX.
- Supported S-O-T-A Algorithms: ResNet, MobileNet v1/v2/v3, SSD, EfficientNet, EfficientDet, YOLOv5, YOLOv7, YOLO8, DeepLabv3, PIDNet and the latest YOLO Models, VLM(CLIP etc.)
Resources:
- Developer page: https://developer.deepx.ai/
- Github page: https://github.com/DEEPX-AI/dx-all-suite
- SDK Install Guide: https://github.com/DEEPX-AI/dx-all-suite/blob/main/docs/source/installation.md
- Clip demo: https://github.com/DEEPX-AI/dx-clip-demo
- OCR demo: https://github.com/DEEPX-AI/DXNN-OCR
We will be back with further updates shortly.
If you are planning to test the DeepX module, see the Digital View boards supporting this:
- ALC-4096-AIH LCD controller and AI: https://www.digitalview.com/controller-alc-4096-aih
- AI-100 board for compute & AI acceleration: https://www.digitalview.com/board-ai-100
