Facial recognition cameras are gaining widespread adoption as they can instantly improve security and customer experience. Faces that are captured in videos can be matched against a database of faces and used differently in different applications in real time.
Avnet has introduced an advanced facial recognition camera based on massive neural networks, enabling a powerful algorithm that runs on a large-scale face database. The use of Deephi's exclusive Deep Compression technology makes it possible to run this powerful neural network on Xilinx's cost-effective FPGAs. Face detection, tracing and quality filtering as well as face feature extraction and comparison can even be supported. What’s more, external linkage devices can be directly controlled through I/O while a built-in supercap protects the data integrity of the TF Card. And with its asymmetric SOC architecture, IPCamera SOC and FPGA, this camera doesn’t just pack a punch, but performs professionally as well.
- Asymmetric SOC architecture: IP Camera SOC + FPGA, powerful and professional.
- Large-scale neural networks training on large scale face database.
- Face detection performance of up to 30fps.Minimum detectable face: 30x30 pixels.
- Face feature extraction performance of up to 20 faces/second.
- Up to 1:30000/1:100000 face pictures match
- Built-in supercap helps to protect data integrity on TF card
- Smart AE based on face detection
- Smart community
- New retail
- Smart shop
- Access control