README.md

metadata

title: Lattice sensAI Generic Object Detection
slug: sensAI-Generic-Object-Detection
short_description: Generic object detection (80 classes)
emoji: 📦
colorFrom: gray
colorTo: yellow
sdk: docker
pinned: false
tags:
  - computer-vision
  - real-time
  - edge-ai
  - on-device
  - embedded
  - low-power
  - object-detection
  - object-detector
  - detector
  - automotive
  - office
  - npu
  - soc
  - fpga
  - sensai

The Lattice sensAI Generic Object Detection demo runs the EVE SDK's MOD (Multi-Object Detection) subsystem against a frame from an uploaded image, an uploaded video, or a live webcam feed. Pick exactly one of three models:

Model	Description
GMOD Base Model	Generic 80-class object detector
Automotive Object Detector	8-class automotive object detector — person, bicycle, car, motorcycle, truck, bus, traffic light, stop sign
Office Object Detector	8-class office-objects detector — bottle, cup, potted plant, laptop, mouse, keyboard, cell phone, book

Each model is a self-contained .tflite loaded through the EVE SDK's custom object-detection model path. GMOD Base Model uses the SDK's bundled weights. Selecting a model in either tab loads it on the worker.

To preview the demo, use the tabs:

• Live Inference — webcam → annotated frames in real time.
• Offline Inference — upload an image or a video; mutually exclusive accordions keep the UI focused on one input at a time.

Note: For demo purposes, the AI pipeline and image-draw operations run on a Hugging Face CPU server. Performance varies with concurrent users.

For model access, visit the Files and versions tab at the top of this page.

For SDK access, fill the form on this page and follow the download instructions. Support: evehelp@latticesemi.com.

About

This demo is maintained by Lattice Semiconductor (LatticeSemi).

License

Proprietary - Lattice Semiconductor Corporation. All rights reserved.

Need a Longer or Commercial License?

• Email: evehelp@latticesemi.com