2026· ITM Web of Conferences· Vol 89, pp. 01006· 0 citations· 5 references
TL;DR
This paper systematically reviews the development of weakly supervised fine-grained image classification, defines its core concepts and task characteristics, and divides it into traditional machine learning and deep learning stages.
Abstract
Fine-grained image classification is a core challenging task in computer vision. It aims to achieve accurate recognition of highly similar subcategories within the same superordinate category. Small inter-class variance and large intra-class variance bring considerable difficulties to feature mining. The weakly supervised paradigm relies only on image-level labels without manual annotation of local features, which greatly reduces annotation costs and has become a mainstream research direction in this field. Taking discriminative feature mining as the core thread, this paper systematically reviews the development of weakly supervised fine-grained image classification, defines its core concepts and task characteristics, and divides it into traditional machine learning and deep learning stages. The performance of three mainstream technical routes is compared based on experiments on classic datasets. Results show that the attention mechanism method is concise and efficient but lacks robustness. The Transformer method delivers better global modeling and generalization ability with high computational cost. The multi-branch innovative network method achieves the best overall performance in accuracy, localization and cross-domain adaptation, representing the optimal current technical route. This paper summarizes existing achievements and limitations, and prospects future directions to provide references for subsequent research.
: Fine-grained visual categorization (FGVC) presents a class of recognition problems in which the discriminative signal is spatially concentrated, visually subtle, and easily destroyed by the preprocessing and augmentation strategies that serve coarse recognition well. Where standard image classification requires a mod...
Richard Adusei, G. Abdul-Salaam· Journal of Artificial Intell...· 0 citations
. Due to the evolution of computer vision, the recognition of fine-grained image classification problems is gaining more and more importance. This paper organizes an overview of the principal conceptualisations of the research and general approaches within the domain of fine-grained image classification, in the view of...
Shen-Gao Wang· Proceedings of the 4th Inter...· 0 citations
Few-Shot Image Classification (FSL) aims to identify and classify new image categories using only a very small number of labeled samples. Addressing the shortcomings of existing meta-learning and metric learning methods, such as insufficient generalization of feature representations with very few samples, overfitting,...
Xua Li, Ru-Jun Wang· International Conference on...· 0 citations
This survey comprehensively evaluates the underlying mechanisms, inherent strengths, and specific weaknesses of each FSIC methods into three categories: Metric Learning, Optimization/Meta-Learning, and Transfer Learning with Large Model Fine-Tuning.
A novel data efficient pyramid vision transformer (DE-PVT), designed to train on limited datasets by utilizing a teacher-student approach and linear computational complexity relative to the number of patches, achieved through a linear spatial reduction mechanism is introduced.
Gazi Jannatul Ferdous, Medhi Hasan Chowdhury, Md. Azad Hossain et al.· Discover Artificial Intellig...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.