Dataset: Total number of images = 810. Training data =675 (80%). Validation data = 135(20%). Number of classes =5 (Worker, Strap, Hardhat, Hook, Harness).
WorkerDetection is trained on dataset annotated with VGG Image annotator
WorkerDetection1 is trained on dataset annotated with Labelme annotator
Integration model is the integration of Mask RCNN and Linknet
Integration file includes the final results (classifying worker as safe or unsafe based on adjacent matrix)