# TAT-VPR: Ternary Adaptive Transformer for Dynamic and Efficient Visual Place Recognition

arXiv: https://arxiv.org/abs/2505.16447 · HTML: https://mlsanity.com/paper/2505.16447/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 13 of its 14 references that are in the map.
- First posted on arXiv 22 May 2025.

## Builds on (most cited first, 13 of 13)

- [ORB-SLAM2: an Open-Source SLAM System for Monocular, Stereo and RGB-D Cameras](https://mlsanity.com/paper/1610.06475/) (2016)
- [Kimera: an Open-Source Library for Real-Time Metric-Semantic Localization and Mapping](https://mlsanity.com/paper/1910.02490/) (2019)
- [Rethinking Visual Geo-localization for Large-Scale Applications](https://mlsanity.com/paper/2204.02287/) (2022)
- [AnyLoc: Towards Universal Visual Place Recognition](https://mlsanity.com/paper/2308.00688/) (2023)
- [MixVPR: Feature Mixing for Visual Place Recognition](https://mlsanity.com/paper/2303.02190/) (2023)
- [Optimal Transport Aggregation for Visual Place Recognition](https://mlsanity.com/paper/2311.15937/) (2023)
- [EigenPlaces: Training Viewpoint Robust Models for Visual Place Recognition](https://mlsanity.com/paper/2308.10832/) (2023)
- [GSV-Cities: Toward Appropriate Supervised Visual Place Recognition](https://mlsanity.com/paper/2210.10239/) (2022)
- [BoQ: A Place is Worth a Bag of Learnable Queries](https://mlsanity.com/paper/2405.07364/) (2024)
- [Q-Sparse: All Large Language Models can be Fully Sparsely-Activated](https://mlsanity.com/paper/2407.10969/) (2024)
- [Binary Neural Networks for Memory-Efficient and Effective Visual Place Recognition in Changing Environments](https://mlsanity.com/paper/2010.00716/) (2020)
- [BitNet a4.8: 4-bit Activations for 1-bit LLMs](https://mlsanity.com/paper/2411.04965/) (2024)
- [Design Space Exploration of Low-Bit Quantized Neural Networks for Visual Place Recognition](https://mlsanity.com/paper/2312.09028/) (2023)

## Abstract (arXiv)

TAT-VPR is a ternary-quantized transformer that brings dynamic accuracy-efficiency trade-offs to visual SLAM loop-closure. By fusing ternary weights with a learned activation-sparsity gate, the model can control computation by up to 40% at run-time without degrading performance (Recall@1). The proposed two-stage distillation pipeline preserves descriptor quality, letting it run on micro-UAV and embedded SLAM stacks while matching state-of-the-art localization accuracy.