# Privacy-Preserving Model Upgrades with Bidirectional Compatible Training in Image Retrieval arXiv: https://arxiv.org/abs/2204.13919 · HTML: https://mlsanity.com/paper/2204.13919/ ## Key facts - Cited by 9 papers in the map (data through 14 Oct 2026). - Builds on 30 of its 40 references that are in the map. - First posted on arXiv 29 Apr 2022. ## Builds on (most cited first, 30 of 30) - [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015) - [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://mlsanity.com/paper/2103.14030/) (2021) - [Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift](https://mlsanity.com/paper/1502.03167/) (2015) - [SGDR: Stochastic Gradient Descent with Warm Restarts](https://mlsanity.com/paper/1608.03983/) (2016) - [Supervised Contrastive Learning](https://mlsanity.com/paper/2004.11362/) (2020) - [CosFace: Large Margin Cosine Loss for Deep Face Recognition](https://mlsanity.com/paper/1801.09414/) (2018) - [In Defense of the Triplet Loss for Person Re-Identification](https://mlsanity.com/paper/1703.07737/) (2017) - [SphereFace: Deep Hypersphere Embedding for Face Recognition](https://mlsanity.com/paper/1704.08063/) (2017) - [Fine-tuning CNN Image Retrieval with No Human Annotation](https://mlsanity.com/paper/1711.02512/) (2017) - [Google Landmarks Dataset v2 -- A Large-Scale Benchmark for Instance-Level Recognition and Retrieval](https://mlsanity.com/paper/2004.01804/) (2020) - [NormFace: L2 Hypersphere Embedding for Face Verification](https://mlsanity.com/paper/1704.06369/) (2017) - [Mutual Mean-Teaching: Pseudo Label Refinery for Unsupervised Domain Adaptation on Person Re-identification](https://mlsanity.com/paper/2001.01526/) (2020) - [Revisiting Oxford and Paris: Large-Scale Image Retrieval Benchmarking](https://mlsanity.com/paper/1803.11285/) (2018) - [Joint Detection and Identification Feature Learning for Person Search](https://mlsanity.com/paper/1604.01850/) (2016) - [Learning with Average Precision: Training Image Retrieval with a Listwise Loss](https://mlsanity.com/paper/1906.07589/) (2019) - [Smooth-AP: Smoothing the Path Towards Large-Scale Image Retrieval](https://mlsanity.com/paper/2007.12163/) (2020) - [Towards Backward-Compatible Representation Learning](https://mlsanity.com/paper/2003.11942/) (2020) - [A Modulation Module for Multi-task Learning with Applications in Image Retrieval](https://mlsanity.com/paper/1807.06708/) (2018) - [Learning Compatible Embeddings](https://mlsanity.com/paper/2108.01958/) (2021) - [Unified Representation Learning for Cross Model Compatibility](https://mlsanity.com/paper/2008.04821/) (2020) - [Asymmetric metric learning for knowledge transfer](https://mlsanity.com/paper/2006.16331/) (2020) - [Towards Universal Backward-Compatible Representation Learning](https://mlsanity.com/paper/2203.01583/) (2022) - [Large-scale Landmark Retrieval/Recognition under a Noisy and Diverse Dataset](https://mlsanity.com/paper/1906.04087/) (2019) - [Self-distillation with Batch Knowledge Ensembling Improves ImageNet Classification](https://mlsanity.com/paper/2104.13298/) (2021) - [End-to-End Deep Kronecker-Product Matching for Person Re-identification](https://mlsanity.com/paper/1807.11182/) (2018) - [Supporting large-scale image recognition with out-of-domain samples](https://mlsanity.com/paper/2010.01650/) (2020) - [Hot-Refresh Model Upgrades with Regression-Alleviating Compatible Training in Image Retrieval](https://mlsanity.com/paper/2201.09724/) (2022) - [1st Place Solution to Google Landmark Retrieval 2020](https://mlsanity.com/paper/2009.05132/) (2020) - [3rd Place Solution to "Google Landmark Retrieval 2020"](https://mlsanity.com/paper/2008.10480/) (2020) - [2nd Place Solution to Google Landmark Retrieval 2021](https://mlsanity.com/paper/2110.04294/) (2021) ## Built on by (most cited first, 9 of 9) - [BT2: Backward-compatible Training with Basis Transformation](https://mlsanity.com/paper/2211.03989/) (2022) - [boldsymbolλ-Orthogonality Regularization for Compatible Representation Learning](https://mlsanity.com/paper/2509.16664/) (2025) - [FastFill: Efficient Compatible Model Update](https://mlsanity.com/paper/2303.04766/) (2023) - [Metric Compatible Training for Online Backfilling in Large-Scale Retrieval](https://mlsanity.com/paper/2301.03767/) (2023) - [Darwinian Model Upgrades: Model Evolving with Selective Compatibility](https://mlsanity.com/paper/2210.06954/) (2022) - [TaCA: Upgrading Your Visual Foundation Model with Task-agnostic Compatible Adapter](https://mlsanity.com/paper/2306.12642/) (2023) - [MixBCT: Towards Self-Adapting Backward-Compatible Training](https://mlsanity.com/paper/2308.06948/) (2023) - [Prototype Perturbation for Relaxing Alignment Constraints in Backward-Compatible Learning](https://mlsanity.com/paper/2503.14824/) (2025) - [Towards Test-time Efficient Visual Place Recognition via Asymmetric Query Processing](https://mlsanity.com/paper/2512.13055/) (2025) ## Abstract (arXiv) The task of privacy-preserving model upgrades in image retrieval desires to reap the benefits of rapidly evolving new models without accessing the raw gallery images. A pioneering work introduced backward-compatible training, where the new model can be directly deployed in a backfill-free manner, i.e., the new query can be directly compared to the old gallery features. Despite a possible solution, its improvement in sequential model upgrades is gradually limited by the fixed and under-quality old gallery embeddings. To this end, we propose a new model upgrade paradigm, termed Bidirectional Compatible Training (BiCT), which will upgrade the old gallery embeddings by forward-compatible training towards the embedding space of the backward-compatible new model. We conduct comprehensive experiments to verify the prominent improvement by BiCT and interestingly observe that the inconspicuous loss weight of backward compatibility actually plays an essential role for both backward and forward retrieval performance. To summarize, we introduce a new and valuable problem named privacy-preserving model upgrades, with a proper solution BiCT. Several intriguing insights are further proposed to get the most out of our method.