Intelligent Vision Brings Artificial Intelligence Directly to the Production Line
With its new AI camera TIV, TURCK promises a paradigm shift in industrial image processing: Instead of complex programming, the intelligent camera only needs to be trained with a few sample images. The TIV (TURCK Intelligent Vision) learns patterns and differences independently and reliably distinguishes good and defective parts or different classes. The training and execution of the neural networks take place directly on the camera, supported by a high-performance 12-megapixel global shutter sensor (Sony Pregius S, 4th generation) and an NVIDIA Jetson Nano GPU with 4 GB of RAM. Image processing runs in real time on the device.
Flexibility and efficiency for practical application
With four pre-installed AI apps – Difference Check, Classifier, Detector, and Code Scan – the TIV12MG-Q110N covers key industrial image processing tasks: from completeness checks and classification to object detection and 1D/2D code recognition. The camera detects and evaluates inspection areas separately, delivering coordinates, confidence scores, and IO/NOI signals directly to PLCs or IT systems. Due to its C-mount connection, standard lenses can be used flexibly; an optional protective housing brings the system up to IP67 protection. The entire control system, lighting technology, sensors, and power supply are decentralized and robust – ideal for direct line placement.
Seamless Integration with TAS
Operation is intuitive via a web browser. Integration into the TURCK Automation Suite (TAS) simplifies device management and integration into digital maintenance and monitoring processes. M12 connectors for power, network, triggers, and I/O allow for flexible integration of the camera into existing systems. On-premise operation, transferable data sets, and neural networks enable scalable rollouts without additional edge hardware or licensing costs.
Key advantages and function:
-
- Quick setup without programming
- Stand-alone in IP67, no additional edge hardware required.
- Real-time results, scalable through transmittable networks
- Difference Check (target-actual comparison via ROIs)
- Classifier & Detector (classification, detection including coordinates and confidence score)
- Code scan (1D/2D, type/position/content), combinable with detector
- High-performance 12 MP sensor, NVIDIA Jetson Nano, C-mount, M12 connectivity
For more information: www.turck.de
