# PoE-Fuse: Precision-Weighted Expert Fusion for Bi-Temporal Change Understanding

arXiv: https://arxiv.org/abs/2609.37485 · HTML: https://mlsanity.com/paper/2609.37485/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 29 of its 40 references that are in the map.
- First posted on arXiv 27 Sep 2026.

## Builds on (most cited first, 29 of 29)

- [Segment Anything](https://mlsanity.com/paper/2304.02643/) (2023)
- [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://mlsanity.com/paper/2301.12597/) (2023)
- [Mamba: Linear-Time Sequence Modeling with Selective State Spaces](https://mlsanity.com/paper/2312.00752/) (2023)
- [Flamingo: a Visual Language Model for Few-Shot Learning](https://mlsanity.com/paper/2204.14198/) (2022)
- [SAM 2: Segment Anything in Images and Videos](https://mlsanity.com/paper/2408.00714/) (2024)
- [What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?](https://mlsanity.com/paper/1703.04977/) (2017)
- [Perceiver: General Perception with Iterative Attention](https://mlsanity.com/paper/2103.03206/) (2021)
- [SAM 3: Segment Anything with Concepts](https://mlsanity.com/paper/2511.16719/) (2025)
- [LISA: Reasoning Segmentation via Large Language Model](https://mlsanity.com/paper/2308.00692/) (2023)
- [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://mlsanity.com/paper/2107.14795/) (2021)
- [RemoteCLIP: A Vision Language Foundation Model for Remote Sensing](https://mlsanity.com/paper/2306.11029/) (2023)
- [GeoChat: Grounded Large Vision-Language Model for Remote Sensing](https://mlsanity.com/paper/2311.15826/) (2023)
- [Fully Convolutional Siamese Networks for Change Detection](https://mlsanity.com/paper/1810.08462/) (2018)
- [Remote Sensing Image Change Detection with Transformers](https://mlsanity.com/paper/2103.00208/) (2021)
- [A Transformer-Based Siamese Network for Change Detection](https://mlsanity.com/paper/2201.01293/) (2022)
- [Multimodal Generative Models for Scalable Weakly-Supervised Learning](https://mlsanity.com/paper/1802.05335/) (2018)
- [Frozen CLIP Models are Efficient Video Learners](https://mlsanity.com/paper/2208.03550/) (2022)
- [SatCLIP: Global, General-Purpose Location Embeddings with Satellite Imagery](https://mlsanity.com/paper/2311.17179/) (2023)
- [ChangeMamba: Remote Sensing Change Detection With Spatiotemporal State Space Model](https://mlsanity.com/paper/2404.03425/) (2024)
- [Mamba-3: Improved Sequence Modeling using State Space Principles](https://mlsanity.com/paper/2603.15569/) (2026)
- [S2Looking: A Satellite Side-Looking Dataset for Building Change Detection](https://mlsanity.com/paper/2107.09244/) (2021)
- [TEOChat: A Large Vision-Language Assistant for Temporal Earth Observation Data](https://mlsanity.com/paper/2410.06234/) (2024)
- [TINYCD: A (Not So) Deep Learning Model For Change Detection](https://mlsanity.com/paper/2207.13159/) (2022)
- [RSCaMa: Remote Sensing Image Change Captioning with State Space Model](https://mlsanity.com/paper/2404.18895/) (2024)
- [RemoteReasoner: Towards Unifying Geospatial Reasoning Workflow](https://mlsanity.com/paper/2507.19280/) (2025)
- [GeoVLM-R1: Reinforcement Fine-Tuning for Improved Remote Sensing Reasoning](https://mlsanity.com/paper/2509.25026/) (2025)
- [BTCChat: Advancing Remote Sensing Bi-temporal Change Captioning with Multimodal Large Language Model](https://mlsanity.com/paper/2509.05895/) (2025)
- [TerraScope: Pixel-Grounded Visual Reasoning for Earth Observation](https://mlsanity.com/paper/2603.19039/) (2026)
- [GeoSolver: Scaling Test-Time Reasoning in Remote Sensing with Fine-Grained Process Supervision](https://mlsanity.com/paper/2603.09551/) (2026)

## Abstract (arXiv)

Bi-temporal change understanding, which localizes and characterizes what changed between two satellite images, is central to disaster response and environmental monitoring, spanning change detection, building localization, and damage assessment. Strong vision-language models address these tasks, but adapting them typically requires full fine-tuning or reinforcement learning, which is costly and unstable. We propose PoE-Fuse, a parameter-efficient framework that instead composes frozen foundation experts for geometry, grounding, and language, resampling their features onto a shared spatial grid and training only a lightweight fusion trunk. PoE-Fuse treats the aligned features as Gaussian observations of a latent scene state and fuses them by learned per-cell precision. This product-of-experts estimator strictly generalizes uniform summation and scalar gating, and extends to change fields by composing the precisions of the two timestamps. A single shared trunk solves the three tasks at once, reaching a mean F1 of 59.2%, compared with 40.7% for an instruction-tuned temporal vision-language assistant, and surpassing dedicated change-detection models retrained under the same protocol and training budget.