| Autor: |
Dornaika, F., Hammoudi, K., Melkemi, M., Phan, T. D. A. |
| Zdroj: |
Signal, Image & Video Processing; Nov2019, Vol. 13 Issue 8, p1611-1617, 7p |
| Abstrakt: |
Efficient analysis of parking slot occupancy is still a complex task in reason of the variety of slot textures and of the difficulty to characterize the relevant information of their associated images. In this paper, we propose a handcrafted approach supported by machine learning techniques. The two main contributions are as follows: Firstly, we introduce a compact handcrafted image descriptor, named pyramid multi-level descriptor (PMLD), designed to capture features at different scales and at different receptive fields in the image region. Secondly, we provide a comparative study of several popular image-based handcrafted and deep learning features. Experiments are conducted on two public datasets: PKLot and CNRPark. It follows that PMLD achieves better results than classical handcrafted descriptors and achieves similar results to those obtained by transfer learning of the deep CNN VGG-F. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
Full text from SpringerLink