Groundlight to developers: build computer vision applications with less programming

Leo Parker Dirac, the founder of Groundlight, believes that using large language models can be expensive and might not always be accurate. He further states that using large language models may not be sufficient for commercial systems. It is essential to involve human judgment in decision-making, especially in handling edge use cases.

“Quality control, process efficiency and continuous improvement are crucial to our success. We’re excited to use Groundlight to inspect our products and monitor our processes without the development overhead of a typical industrial solution,” says Uriel Eisen, the founder of Austere Manufacturing.

Groundlight released an open-source project for its image processing service that works with an M5Stack ESP32 camera. This technology utilizes a camera to capture images and sends them to the Groundlight API via Wi-Fi. It will then send a notification on Slack once it receives a positive response, followed by a change in state.

The computer vision system can analyze images using machine learning models even if no pre-existing data is available to be labeled. If the model is confident about the analysis, it answers. Intensive analysis methods are used if uncertain until a human reviews the image in real-time.

The image processing service of Groundlight may experience delays because humans need to review the image queries. However, the company believes the computer vision system will improve with time to provide more accurate and faster responses.

“Groundlight’s approach enables companies to utilize off-the-shelf cameras and inexpensive equipment to quickly build and reliably operate customized models. We are excited to work with this brilliant team as they build the company,” says Tim Porter, the managing director of Madrona.

Article Topics

AI/ML  |  computer vision  |  DevOps  |  edge AI  |  Groundlight  |  model