Image classification based on The Combination of Improved Bag of visual words and deep learning techniques
| Main Author: | Marwa. A. Marzouk* |
|---|---|
| Format: | Article Journal |
| Terbitan: |
, 2022
|
| Subjects: | |
| Online Access: |
https://zenodo.org/record/6379851 |
Daftar Isi:
- Abstract: Traditional images classification methods are frequently utilized in practical applications, but they have many disadvantages, including disappointing effects, low classification accuracy, and a lack of adaptability. In this paper, we propose an image classification model by the Integration of Convolution Neural Network and modified Bag of Visual Word (MBoVW). A BoVW is capable of picking features in image classification; however, it is only useful if the feature extractor provided is well-matched. in this paper image classification, BoVW uses Scale- Invariant Feature Transform (SIFT) and Oriented Fast and Rotated BRIEF (ORB) descriptors as a result, the SIFT-ORB-BoVW model that was created has highly discriminant features, which improves classification performance. We've also looked into using a fuzzy bag of visual words to find relevant photos and solve the problem of visual code words being hard to assign to local image properties. Adding fuzziness to an image improves its qualities. Convolutional neural networks (CNNs)/ Support vector machines (SVM) are also used in this paper to increase the suggested feature extractor's potential to learn more meaningful visual vocabulary from the image. In different datasets containing a variety of image categories, our method outperforms traditional BoVW as a tool for image classification.