Power-Efficient Object Detection in PIR Data Using Syntiant NDP

Successfully achieved two objectives including accurate classification based on motion sensor data-based movement and detecting objects in passive infrared images • Created four top-performing deep learning models (Random Forest, Decision Tree, LGMB, KNN) by utilizing the PyCaret library. Using the grid search method, evaluated the best model (Random Forest Classifier) with an accuracy score of 0.8931 • Utilized YoloV3, YoloV5, and OpenCV with DNN_Detection methods for object detection. Among all, YoloV5 had the best performance with an average of 0.83