Research map / Label-efficient and robust learning / Domain adaptation and generalization
WildIng: A Wildlife Image Invariant Representation Model for Geographical Domain Shift
Julian D. Santamaria, Claudia Isaza, Jhony H. Giraldo
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 LLM descriptions to image plus image-text encoders (ablation) | WildIng · Terra Incognita (out-of-domain, trained on Snapshot Serengeti) | Cis-Test accuracy: 47.62 -> 50.06 | Improved |
| add LLM descriptions to image plus image-text encoders (ablation) | WildIng · Terra Incognita (out-of-domain, trained on Snapshot Serengeti) | Trans-Test accuracy: 38.96 -> 39.96 | Improved |
| add LLM descriptions to image encoder only (ablation) | WildIng · Terra Incognita (out-of-domain, trained on Snapshot Serengeti) | Cis-Test accuracy: 46.67 -> 40.32 | Hurt |
| add LLM descriptions to image encoder only (ablation) | WildIng · Terra Incognita (out-of-domain, trained on Snapshot Serengeti) | Trans-Test accuracy: 37.18 -> 39.11 | Improved |
| add image encoder to image-text encoder (ablation) | WildIng · Terra Incognita (out-of-domain, trained on Snapshot Serengeti) | Cis-Test accuracy: 28.14 -> 47.62 | Improved |
| add image encoder to image-text encoder plus LLM (ablation) | WildIng · Terra Incognita (out-of-domain, trained on Snapshot Serengeti) | Trans-Test accuracy: 26.66 -> 39.96 | Improved |
| add image-text encoder to image encoder (ablation) | WildIng · Terra Incognita (out-of-domain, trained on Snapshot Serengeti) | Cis-Test accuracy: 46.67 -> 47.62 | Improved |
| use ChatGPT descriptions instead of LLAMA (ablation) | WildIng · Terra Incognita (out-of-domain, trained on Snapshot Serengeti) | Cis-Test accuracy: 28.82 -> 50.06 | Improved |
| use ChatGPT descriptions instead of Qwen (ablation) | WildIng · Terra Incognita (out-of-domain, trained on Snapshot Serengeti) | Trans-Test accuracy: 22.88 -> 39.96 | Improved |
| use ChatGPT descriptions instead of Phi (ablation) | WildIng · Terra Incognita (out-of-domain, trained on Snapshot Serengeti) | Trans-Test accuracy: 19.96 -> 39.96 | Improved |
| increase number of short descriptions per class (ablation) | WildIng · Terra Incognita (out-of-domain, trained on Snapshot Serengeti) | accuracy | Improved |
| add adapter to CLIP | CLIP · Terra Incognita (out-of-domain, trained on Snapshot Serengeti) | Cis-Test accuracy: 39.14 -> 27.45 | Hurt |
Closest papers by meaning
- WildCLIP: Scene and Animal Attribute Retrieval from Camera Trap Data with Domain-Adapted Vision-Language Models (IJCV 2024 · similarity 0.91)
- Reconstructing Animals and the Wild (CVPR 2025 · similarity 0.88)
- Generating the Wild: Individual-Consistent Image-to-Video Generation for Wildlife (NeurIPS 2026 · similarity 0.87)
- MammAlps: A Multi-view Video Behavior Monitoring Dataset of Wild Mammals in the Swiss Alps (CVPR 2025 · similarity 0.86)
- DivShift: Exploring Domain-Specific Distribution Shifts in Large-Scale, Volunteer-Collected Biodiversity Datasets (AAAI 2025 · similarity 0.86)
- The PanAf-FGBG Dataset: Understanding the Impact of Backgrounds in Wildlife Behaviour Recognition (CVPR 2025 · similarity 0.85)
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.
