English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Since neural networks are data-hungry, incorporating data augmentation intraining is a widely adopted technique that enlarges datasets and improvesgeneralization. On the other hand, aggregating predictions of multipleaugmented samples (i.e., test-time augmentation) could boost performance evenfurther. In the context of person re-identification models, it is commonpractice to extract embeddings for both the original images and theirhorizontally flipped variants. The final representation is the mean of theaforementioned feature vectors. However, such scheme results in a gap betweentraining and inference, i.e., the mean feature vectors calculated in inferenceare not part of the training pipeline. In this study, we devise the FlipReIDstructure with the flipping loss to address this issue. More specifically,models using the FlipReID structure are trained on the original images and theflipped images simultaneously, and incorporating the flipping loss minimizesthe mean squared error between feature vectors of corresponding image pairs.Extensive experiments show that our method brings consistent improvements. Inparticular, we set a new record for MSMT17 which is the largest personre-identification dataset. The source code is available athttps://github.com/nixingyang/FlipReID.

FlipReID: Closing the Gap between Training and Inference in Person Re-Identification