Hanwha Vision is launchings its AI-powered System on a Chip (SoC), Wisenet 9. After nearly four years of research and testing, the manufacturer says Wisenet 9โs features help elevate its surveillance cameras into an AI-driven era.
John Lutz Boorman, Head of Product and Marketing at Hanwha Vision Europe, pictured, said: โWith Wisenet 9, you can count on the superior image quality that delivers the best possible video in the toughest environments, as well as the accuracy of the AI-generated data, whether statistical or event-driven.”
The developers say that AI lets systems analyse vast amounts of video data and discern crucial patterns and anomalies. A differentiator the firm points to is the two Neural Processing Units (NPUs), which the firm says have improved inference performance three-fold compared to Wisenet 7, the previous SoC generation.
While one NPU handles image processing, the other focuses on object detection and analytics. This dual NPU concept was introduced to see that video quality and analytics have independent resources, preventing one function from impacting the performance of the other. Wisenet 9 uses AI to analyse visual data with precision. By leveraging a NPU, the system moves past traditional image processing to extract critical information, unlocking previously unseen details with clarity the developers say.
This visual intelligence is achieved through AI features, including AI-based extreme Wide Dynamic Range, AI-powered noise reduction, alongside image enhancements. The AI-driven image enhancement, powered by an independent NPU, ensures that every detail, no matter how subtle, is captured and analysed, the makers add. By intelligently reducing ‘noise’, Wisenet 9 achieves reduced bandwidth consumption and storage, the company says, using H.265 compression of video. The H.265 codec provides video quality with optimised bandwidth, the firm says, giving real-time video processing and transmission.
Wisenet 9โs H.265 codec, aligned with industry standards and ONVIF profiles, ensures hardware compatibility and integration with security equipment. As for AI for analytics, you can use attribute analysis, including colour, bag, face mask, glasses, age, and gender, with other key analytics such as people and vehicle counting, slip and fall detection, crowd detection, heat map, and queue management.
Wisenet 9 incorporates also:
Re-Identification (RE-ID) for person tracking and searching:
RE-ID technology is a deep learning-based approach to matching visual identities across non-overlapping camera views. RE-ID extracts feature vectors representing visual attributes, enabling video management systems to recognise and track individuals even with changes in appearance or conditions. For example, it can identify a person entering a building with an orange jacket and later recognise the same person leaving with the jacket unzipped and a hat, across different cameras.
Dynamic Privacy Masking (DPM):
This AI-based feature safeguards biometric personal data, ensuring compliance with privacy regulations. Wisenet 9 offers masking options, including opaque and mosaic masking, allowing users to tailor privacy protection
AI Packs:
Wisenet 9 is compatible with AI Packs (Traffic, Retail, Factory), designed to deliver actionable insights specific to various sectors.
AI Sound Classification:
AI-based sound classification enables detection and directional awareness of critical events like gunshots, glass breaks, and screaming, for situational awareness.
Hanwha Vision Open Platform:
Wisenet 9 lets users customise security with applications for video analytics, storage, and other functionalities. As for end-to-end protection, from Wisenet 7 it inherits features like Secure Storage, OS, and Boot, which protect critical chipset data and prevent unauthorised software execution. The inclusion of a FIPS 140-3 Level 3 means that the chipset adheres to security protocols, assuring users of data protection. Visit HanwhaVision.eu/.
