list of paper on vehicle reid
Published:
- Group-Sensitive Triplet Embedding for Vehicle Reidentification
- diverse but limited data -> modeling of appearance variance of each car
- large scale of cars
- training objective: min intra-class dist. s.t. inter-class dist. > threshold
- triplet loss: class center & group center
- Vehicle Re-identification by Fusing Multiple Deep Neural Networks
- siamese on Alex, VGG, FRCNN feature map
- Deep Relative Distance Learning: Tell the Difference Between Similar Vehicles
- same as the first paper
- diff. fc head for attribute recog. & embedding
- special marks for identification
- Unsupervised Vehicle Re-Identification using Triplet Networks
- detector and tracker -> feature extractor -> matching
- require video data
- Viewpoint-aware Attentive Multi-view Inference for Vehicle Re-identification
- 5 views for a car: front, side, rear, and two in the middle
- input image -> feature -attention-> 5 map of diff. area masks -> discriminator v.s. real images
- how realistic is generated masks?
- attention? central point feature in each vp cluster
- Learning Coarse-to-Fine Structured Feature Embedding for Vehicle Re-Identification
- fine-tune using images from the same vehicle