Weakly Supervised Fine-Grained Image Classification via Salient Region Localization and Different Layer Feature Fusion

Abstract

<jats:p>The fine-grained image classification task is about differentiating between different object classes. The difficulties of the task are large intra-class variance and small inter-class variance. For this reason, improving models’ accuracies on the task heavily relies on discriminative parts’ annotations and regional parts’ annotations. Such delicate annotations’ dependency causes the restriction on models’ practicability. To tackle this issue, a saliency module based on a weakly supervised fine-grained image classification model is proposed by this article. Through our salient region localization module, the proposed model can localize essential regional parts with the use of saliency maps, while only image class annotations are provided. Besides, the bilinear attention module can improve the performance on feature extraction by using higher- and lower-level layers of the network to fuse regional features with global features. With the application of the bilinear attention architecture, we propose the different layer feature fusion module to improve the expression ability of model features. We tested and verified our model on public datasets released specifically for fine-grained image classification. The results of our test show that our proposed model can achieve close to state-of-the-art classification performance on various datasets, while only the least training data are provided. Such a result indicates that the practicality of our model is incredibly improved since fine-grained image datasets are expensive.</jats:p>

Development of Effective Methods for Structural Image Recognition Using the Principles of Data Granulation and Apparatus of Fuzzy Logic

“…The methodological aspects of structural image recognition are presented in [9], [10], [15]- [19]. The papers mentioned above [9], [10], [15]- [19] does not contain the consideration of the fuzziness, incompleteness, redundancy and inconsistency of input data during the development of modern information systems. There is also no practical implementation of the proposed method of structural image recognition in [10], [11].…”

“…The performance indicator of recognition due to the transformation of the system of features can be improved tenfold [15], [42]. For example, granulation of description elements [36] can provide:…”

“…In the methods of structural image recognition, the individual keypoints of description elements are represented by vectors from space B n with real components. The structural description of the object is a finite multiset O ⊂ B n [15].…”

“…The SURF method is used to construct descriptions of objects for the following tasks: comparison of images [30]; search for objects on images [15]; 3D reconstruction of the territory infrastructure [20]; recognition of road signs [45]; automatic recognition of bacteria in blood tests [41], etc.…”

“…The SURF description can contain hundreds of vectors, that significantly slows down processing [15]. The problem of reducing the number of vectors involves the construction of a concise description O * on the base of the image…”