Hikvision smart thermal bullet cameras with increased sensitivity


Hikvision has launched a new thermal bullet camera (DS-2TD2117) that incorporates deep learning capabilities and a new sensor designed to improve the quality of its images. This product brings thermal technology cameras to a lower budget scenario, making it practical for smaller solutions.

Developing on earlier versions, the Deepinview camera is equipped with a new improved 160 x 120 pixel sensor in the Hikvision family. The camera is also more sensitive – detecting to a noise-equivalent temperature difference (NETD) of 40mk (previous models went to 50mk). NETD is an industry standard measurement of a sensor’s sensitivity to thermal radiation in certain parts of the electromagnetic spectrum.

Based on deep learning algorithms, the camera is also said to deliver powerful and accurate behaviour analysis, including detections such as line crossing, intrusion, region entrance and exit. The intelligent human/vehicle detection feature helps reduce false alarms caused by animals, camera shake, falling leaves, or other irrelevant objects, significantly improving alarm accuracy. In addition, the cameras are equipped with a built-in GPU. This high-performance GPU can support updates with more complex algorithms with larger data samples in the future to further improve the intelligent effect of video content analytics (VCA).

Other features include temperature exception alarm functionality, advanced fire detection and smoke detection algorithm, support for contrast adjustment, shutter adjustment in various modes, and 3D DNR. The camera’s short focal length (3mm or 6mm) makes it a crucial element in smaller perimeter solutions, like yards or car parks for small businesses. Its fire protection capabilities come in handy in solutions like warehouses and rubbish dumps, where avoiding fire is a priority. In essence, this camera can help reduce costs, and prevent damage but also, more importantly, can increase safety by detecting potential fires before they happen.