| Autor: |
Majuran, Shajini, Ramanan, Amirthalingam |
| Předmět: |
|
| Zdroj: |
Visual Computer; Dec2023, Vol. 39 Issue 12, p6609-6623, 15p |
| Abstrakt: |
Fashion is defined as a prevailing custom or style of dress, etiquette, and socialising. In recent years, fashion clothing analysis has attracted extensive attention from many researchers due to the introduction of large-scale datasets and the use of deep learning techniques. In this work, we propose a single-stage attention-based network for fashion clothing detection and classification. The proposed network is a single-stage detection which benefits from adopting multidimensional features through a multilevel architecture, so that the semantic gap between the lower- and upper-level features from different levels of feature representation is resolved. Besides, the network is structured based on multilevel contextual features retrieved using attention blocks in a global manner that implements a strong visual attention. Further, the classification and detection branches maintain fewer trainable parameters; thus, the model not only shows efficiency but also the testing results show state-of-the-art performance in fashion clothing detection and classification evaluated on large-scale DeepFashion2 dataset. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
Full text from SpringerLink