Algorithm

Our advanced IP camera solution delivers:

✅ 100,000-face database capacity
✅ 50-person real-time recognition
✅ Perfect for school attendance tracking & security monitoring

🔹 Why Choose FengCam?

• Self-developed AI algorithms for unmatched accuracy
• High-performance hardware built for demanding scenarios
• Proven solutions in education & security sectors

Face detection is one of the most important visual tasks. With our self-developed CNN (Convolutional Neural Network) detection algorithm, together with the multi-scene face detection datasets we have accumulated over the years, we have achieved sota-level detection accuracy on various open-source test datasets. The lightweight detection model requires only 0.5Tops of computing power to achieve real-time detection frame rate (30fps). It can be used in various face-related tasks such as people counting, facial recognition, and facial attributes.

High-precision and lightweight person body detection model, together with self-developed person body posture detection algorithm, can achieve real-time person body posture detection for 8 people under only 1Tops computing power, and the algorithmic datasets of the whole scenario can be used for intelligent care, fall detection, forbidden area detection, electronic fence and other applications.

With only 1Tops of computing power, it can perform real-time vehicle detection, vehicle color recognition, vehicle logo recognition, and license plate recognition, which can be used in parking lots, highways, smart cities, and other scenarios.

The high-precision and lightweight vehicle detection model, together with the self-developed tracking algorithm, makes it possible to realize vehicle attributes (vehicle colour, vehicle brand, vehicle direction of travel, etc.), number plate recognition and other tasks while vehicle tracking.

Highly accurate and lightweight pedestrian detection model with attached output nodes for key points of faces, together with self-developed tracking algorithms, makes it possible to efficiently detect faces while pedestrian tracking for tasks such as face localization and facial recognition.