ДОСЛІДЖЕННЯ ТА ЗАСТОСУВАННЯ МЕТОДІВ COMPUTER VISION ДЛЯ РОЗПІЗНАВАННЯ ТА КЛАСИФІКАЦІЇ ЗАБРУДНЕНЬ

The pressing issue of environmental preservation has prompted global efforts to minimize human impact on the planet. This research focuses on leveraging Computer Vision, specifically CNNs, for waste recognition on mobile devices. The study uses the TACO dataset, resembling real-world scenarios. Data augmentation is applied to address limited training data, doubling input effectiveness. The goal is a CNN model capable of efficiently recognizing and classifying common waste types in real-time on mobile devices. Challenges include meticulous data preprocessing, training computational costs, model selection, porting to mobile devices, and real-world cases assessment. Embracing AI technologies can enhance human efficiency in environmental preservation, fostering a more sustainable and comfortable existence.