Cinterion, a business unit of Telit, augments its device WISE AI Visual Inspection by integrating NVIDIA's latest TAO 6.0 technology.
In a significant development, Telit Cinterion has announced the integration of its deviceWISE AI Visual Inspection platform with NVIDIA TAO 6.0. This integration aims to revolutionise industrial visual inspection tasks by improving accuracy and efficiency.
Key Features
The integration brings advanced annotation tools such as prompt segmentation and in-context segmentations. These tools enable auto-labeling of object detection and segmentation masks using simple text prompts, reducing the need for extensive training or fine-tuning.
The integration also utilises pre-trained foundation models like FoundationPose for estimating object poses, alongside other models such as NV-DINOv2, NV-CLIP, and GroundingDINO. These models are fine-tuned with minimal custom data via transfer learning, supporting multi-modal and delivering high-performance inferencing.
Moreover, the integration provides low-code auto-labeling capabilities, allowing for faster deployment without the need for custom code or specialty cameras. It also supports multi-modal operations, enhancing the versatility of the inspection platform.
Benefits
Manufacturers can build and deploy custom AI models with unparalleled ease and speed, thanks to the integration with NVIDIA TAO 6.0. The use of advanced annotation tools and pre-trained models improves accuracy for specific industrial objects while reducing annotation costs and effort.
The integration ensures seamless on-premises operation, maintaining robust security for proprietary data without cloud dependencies. It enables manufacturers to collect, transform, and integrate data from any machine to any IT system, supporting full applications for Industry 4.0 and digital transformation.
With TAO's low-code auto-labelling, pose-estimation, and multi-modal models built in, customers get faster deployment, higher accuracy, and zero custom code or specialty cameras. The platform is a turnkey solution for various industries, including automotive, pharmaceutical, healthcare, and energy.
For more information about deviceWISE AI Visual Inspection with NVIDIA TAO, visit telit.com/devicewise-ai-consultation/.
Using the advanced annotation tools in the integration, such as prompt segmentation and in-context segmentations, allows for the auto-labeling of object detection and segmentation masks, reducing the extensive training required for AI models. The integration also utilizes pre-trained foundation models like FoundationPose for estimating object poses, alongside other models like NV-DINOv2, NV-CLIP, and GroundingDINO, to deliver high-performance inferencing and improved accuracy in artificial-intelligence applications within various industries.
