Performance Evaluation between RGB and YCrCb in TC-SF-CNNLS for vehicle type recognition system
In this paper, the performance impact of vehicle type recognition (VTR) using Three-Channel of Sparse-Filtered Convolutional Neural Network with Layer-Skipping strategy (TC-SF-CNNLS) technique is observed. Unlike other techniques that lacking in extracting unique features, TC-SF-CNNLS able to extrac...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Conference or Workshop Item |
| Language: | en |
| Published: |
Industrial Engineering and Applications Organization
2021
|
| Subjects: | |
| Online Access: | https://umpir.ump.edu.my/id/eprint/32727/1/27.%20Performance%20Evaluation%20between%20RGB%20and%20YCrCb%20in%20TC-SF-CNNLS%20for%20vehicle%20type%20recognition%20system.pdf https://umpir.ump.edu.my/id/eprint/32727/ https://doi.org/10.1109/ICIEA52957.2021.9436723 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In this paper, the performance impact of vehicle type recognition (VTR) using Three-Channel of Sparse-Filtered Convolutional Neural Network with Layer-Skipping strategy (TC-SF-CNNLS) technique is observed. Unlike other techniques that lacking in extracting unique features, TC-SF-CNNLS able to extract the unique features from colour image. However, colour image can be in many schemes. Thus, we implemented the technique with two colour schemes which is RGB and YCrCb to analyze which one is able to give better performance in the VTR. We tested the implementation with a benchmark dataset known as BIT and self-obtained dataset known as SPINT. The results are observed based on the accuracy performance that represented in a confusion table. Based on the results, YCrCb outperformed RGB with the highest average accuracy 90.5% and 90.4% for both datasets, respectively. We can conclude that TC-SF-CNNLS is best performed when YCrCb is used as the colour scheme in extracting the vehicle features from the colour image. |
|---|