This solution could process live stream or local files on its 8-channels simultaneously with 30fps/channel. The in-built H.264/H.264 codec & hardware scaler converts the streams feed them to the AI engine for processing. AI engine (DPU) deployed on FPGA which will extract features up to 2622 points for the facial analysis process.
Corazon-AI provides the flexibility to implement the pre/post-processing operations on the FPGA in order to achieve the 30fps/channel. On the energy consumption side, Corazon-AI consumes optimized power 1.5W/stream to process the inference. Corazon-AI can store the processed data in a local database as well as cloud transfer via ZMQ, RabbitMQ protocols.