Features:
- 4 MP Resolution
- DarkFighter Low-Light Performance
- 140 dB WDR for Clear Backlit Images
- H.265+ Compression
- Smooth Live View and Data Self-Correcting
- 5 Streams + Up to 5 Custom Streams
- Vandal-Proof (IK10)
Attribute-Based People Counting Embedded deep learning algorithms integrate multiple intelligences. They count people, sample face features, and identify people who pass repeatedly over a period of time. They count persons and send a face alarm for entrance control and people counting. Hard Hat Detection Embedded deep learning algorithms detect persons in a specified region. They detect if the person is wearing a hard hat and captures the person's head and sends an alarm if not. Perimeter Protection The camera performs perimeter protection, monitoring line crossing, intrusion, region entrance, and region exiting. Algorithms filter out mistaken alarms caused by leaves, lights, animals, flags, etc.
Multi-Target-Type Detection Deep learning algorithms detect and capture faces, human bodies, and vehicles in specified regions and outputs features such as gender, age, top color, and vehicle type. It supports a structurized modeling of the face and human body to achieve structurized data collection. Queue Management Embedded deep learning based algorithms detect queued people number and waiting times. The body feature detection algorithm filters out wrong targets and increases detection accuracy. Metadata Metadata uses individual application data instances or the data content. Metadata can be used for third-party application development. Smooth Streaming Smooth streaming video quality in different network conditions. For example, in poor network conditions, streaming bit rate and resolution are automatically adjusted to avoid mosaic and lower latency in live view; in multiplayer network conditions, the camera transmits redundant data for self-error correcting in back-end device to solve mosaic problems because of packet loss and error rate.