Research map / Time series / Long-term forecasting
TAMMs: Change Understanding and Forecasting in Satellite Image Time Series with Temporal-Aware Multimodal Models
Zhongbin Guo, Yuhao Wang, Ping Jian, Chengzhi Li, Xinyue Chen, Zhen Yang, Ertai E
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Physical Time Encoder (time-interval token) (ablation) | TAMMs · Temporal change description (150-sequence Google Earth test set) | TCS: 0.7456 -> 0.9690 | Improved |
| add Contextual Temporal Prompting (ablation) | TAMMs · Temporal change description (150-sequence Google Earth test set) | TCS: 0.8234 -> 0.9690 | Improved |
| add PTE and CTP over plain SFT of the MLLM (ablation) | TAMMs · Temporal change description (150-sequence Google Earth test set) | CIDEr-D: 0.7523 -> 0.9030 | Improved |
| add semantic feature fusion (SFCI) into diffusion control (ablation) | TAMMs · Future satellite image forecasting (150-sequence Google Earth test set) | TCS: 0.7911 -> 0.9690 | Improved |
| add text guidance alongside semantic fusion (ablation) | TAMMs · Future satellite image forecasting (150-sequence Google Earth test set) | TCS: 0.9410 -> 0.9690 | Improved |
| add semantic guidance of both kinds over base control block (ablation) | TAMMs · Future satellite image forecasting (150-sequence Google Earth test set) | TCS: 0.7624 -> 0.9690 | Improved |
| add semantic fusion on SSIM (ablation) | TAMMs · Future satellite image forecasting (150-sequence Google Earth test set) | SSIM: 0.1596 -> 0.1831 | Improved |
| vary the CTP prompt (short, default, verbose) (ablation) | TAMMs · Future satellite image forecasting (150-sequence Google Earth test set) | TCS: 0.9580 / 0.9690 / 0.9615 | No clear effect |
| replace native unified models (Chameleon, Janus-Pro) with MLLM plus diffusion pairing | TAMMs · remote sensing tasks | quality | Improved |
| use the unified framework vs specialist forecasting baselines | TAMMs · Future satellite image forecasting (150-sequence Google Earth test set) | SSIM: 0.1831 vs 0.2098 (MCVD) | Hurt |
Closest papers by meaning
- TEOChat: A Large Vision-Language Assistant for Temporal Earth Observation Data (ICLR 2025 · similarity 0.84)
- DiffusionSat: A Generative Foundation Model for Satellite Imagery (ICLR 2024 · similarity 0.83)
- DynamicVL: Benchmarking Multimodal Large Language Models for Dynamic City Understanding (NeurIPS 2025 · similarity 0.83)
- Terra: A Multimodal Spatio-Temporal Dataset Spanning the Earth (NeurIPS 2024 · similarity 0.83)
- Causal Deciphering and Inpainting in Spatio-Temporal Dynamics via Diffusion Model (NeurIPS 2024 · similarity 0.83)
- The Change You Want To Detect: Semantic Change Detection In Earth Observation With Hybrid Data Generationf (CVPR 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
