Comparison of Garbage Classification Frameworks Using Transfer Learning and CNN

Literature Review

Before the development in the field of deep learning models, a big chunk of features was delineated manually (LeCun et al., 1989) and then provided to the models. Many Deep CNN models have been introduced for the image classification process (Vo et al., 2018; Vo et al., 2019; Hu et al., 2019) like RR-VGG, HF-SD (Hybrid Framework for Smile Detection), CS-LSMP, etc. Apart from these CNN models based on image processing, machine learning has built a stand-in early diagnosis and classification processes in various industries (Sahu et al., 2020), medical (Gourisaria et al., 2021), agricultural, etc. models. Yang et al. (Yang and Thung, 2016) provided a Trashnet dataset that comprises six different classes namely glass, paper & cardboard, plastic, metal, and trash. All the data samples were of image type which was captured using a variety of devices.