Emotion Analyzer

Emotional marketing for retail, services, and more. Training was conducted on the well-known dataset Expression in-the-Wild (ExpW). The number of classes - 7. The total number of images is 91,793. Training was carried out on Google Colaboratory using the library. Model architecture - YOLO11; Inference implemented in two variants in Python using the Ultralytics library, and in C++ using TensorRT for NVIDIA graphics cards, and for mini computers NVIDIA Jetson Nano. Two models are used for implementation. The first is face inference, the second is emotion inference, for NVIDIA models YOLO converted into engine engines. Ultralytics.