Analysis on Fine-Grained Image Recognition Datasets and Techniques: A Review

The straightforward task of working with objects that fall under several intra-class categories inside the same inter-class category group, such as various bird species or various motor vehicle types, is the subject of FGIR (Fine-Grained Image Recognition). FGIR can find applications in crop disease detection, satellite image detection, and several other use cases in biological research. FGIR is made up of instance-level analysis and basic-level category analysis. The objective is to look at visual components from subordinate categories, such as feathered creature species or diverse car demonstration types. We have studied several methodologies and techniques used in image recognition that can also be used in FGIR. We have also studied several datasets for image recognition and some for FGIR use solely. After analysing recent developments in the field, we have isolated and summarised some of the techniques used. These divisions are based on the image recognition models and picture preprocessing methods employed.