Abstract:
Vehicle re-identification (Re-ID) research has grown due to intelligent traffic control and monitoring. Vehicle Re-ID is more difficult and unpredictable than person Re-ID because viewpoint variations greatly affect vehicle appearance. Existing studies mostly extract global features based on visual appearance to represent the target vehicle, but viewpoint variation is rarely considered. To improve vehicle Re-ID, we introduce latent view labels by clustering and use view information in deep metric learning. We also tighten the center constraint to improve feature space intra-class compactness. To improve vehicle separability, we use orthogonal regularization. VARID outperforms state-of-the-art with 79.3% mAP on VeRi-776 and 88.5% on VehicleID. The proposed method outperforms state-of-the-art methods on four benchmarks.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
