Image Classification Training and Annotation - BinSight Project
I trained an image classification model using CreateML on a dataset of over 10,000 images. The project involved manually curating and segmenting images across 12 classes to improve model performance. Extensive preprocessing and evaluation steps were performed to reach a high accuracy level and on-device efficiency. • Created and labeled a diverse image dataset for waste classification. • Applied image segmentation techniques to enhance annotation quality. • Trained, evaluated, and iteratively refined the CreateML image classifier. • Validated model performance with rigorous testing and result analysis.