Revolutionizing Emotional Intelligence: AI-Driven Mood and Pose Detection through Facial Data

Pose detection and mood recognition have always been at the heart of psychology, human computer interface, and security. This paper forms an approach toward robust and efficient human pose and face detection in video streams spanning across domains in a real-time approach. This can be done through the sequentially pipelined approach that applies MediaPipe Pose model for robust landmark detection and HAAR Cascade Classifier for identifying the precise face from video frames. The proposed system tested in a series under many different settings and analyses. According to this study, as much as pose detection is concerned, the results are way more accurate in lesser distances, which means distance ranges from 0.5 meters to reach almost perfect accuracy, to 98 percent up to 2 meters, then a reduction of almost 90 percent. It found an accuracy range in mood to be roughly around 99 percent correct, except for 1 percent accuracy across all tested in the Neutral at a distance of 1 m. This suggests that even though the system is excellent at emotion recognition, its performance in the detection of poses at more than 1 meter and in mood recognition when presented at a distance of 1 meter is very low.