Content-Based Image Retrieval Using Hybrid Densenet121-Bilstm and Harris Hawks Optimization Algorithm
| Autor: | null Sanjeevaiah K., Tatireddy Subba Reddy, Sajja Karthik, Mahesh Kumar, null Vivek D. |
|---|---|
| Rok vydání: | 2022 |
| Předmět: | |
| Zdroj: | International Journal of Software Innovation. 11:1-15 |
| ISSN: | 2166-7179 2166-7160 |
| DOI: | 10.4018/ijsi.315661 |
| Popis: | In the field of digital data management, content-based image retrieval (CBIR) has become one of the most important research areas, and it is used in many fields. This system searches a database of images to retrieve most visually comparable photos to a query image. It is based on features derived directly from the image data, rather than on keywords or annotations. Currently, deep learning approaches have demonstrated a strong interest in picture recognition, particularly in extracting information about the features of the image. Therefore, a Densenet-121 is employed in this work to extract high-level and deep characteristics from the images. Afterwards, the training images are retrieved from the dataset and compared to the query image using a Bidirectional LSTM (BiLSTM) classifier to obtain the relevant images. The investigations are conducted using a publicly available dataset named Corel, and the f-measure, recall, and precision metrics are used for performance assessment. Investigation outcomes show that the proposed technique outperforms the existing image retrieval techniques. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|