README.md

---
license: apache-2.0
dataset_info:
- config_name: clip_quad_modal
  features:
  - name: 'Unnamed: 0.4'
    dtype: int64
  - name: image_url
    dtype: string
  - name: sound_url
    dtype: string
  - name: 'Unnamed: 0.3'
    dtype: int64
  - name: 'Unnamed: 0.2'
    dtype: int64
  - name: 'Unnamed: 0.1'
    dtype: int64
  - name: 'Unnamed: 0'
    dtype: int64
  - name: id
    dtype: int64
  - name: observed_on_string
    dtype: string
  - name: observed_on
    dtype: string
  - name: time_observed_at
    dtype: string
  - name: time_zone
    dtype: string
  - name: user_id
    dtype: int64
  - name: user_login
    dtype: string
  - name: user_name
    dtype: string
  - name: created_at
    dtype: string
  - name: updated_at
    dtype: string
  - name: quality_grade
    dtype: string
  - name: license
    dtype: string
  - name: url
    dtype: string
  - name: tag_list
    dtype: string
  - name: description
    dtype: string
  - name: num_identification_agreements
    dtype: int64
  - name: num_identification_disagreements
    dtype: int64
  - name: captive_cultivated
    dtype: bool
  - name: oauth_application_id
    dtype: float64
  - name: place_guess
    dtype: string
  - name: latitude
    dtype: float64
  - name: longitude
    dtype: float64
  - name: positional_accuracy
    dtype: float64
  - name: private_place_guess
    dtype: float64
  - name: private_latitude
    dtype: float64
  - name: private_longitude
    dtype: float64
  - name: public_positional_accuracy
    dtype: float64
  - name: geoprivacy
    dtype: string
  - name: taxon_geoprivacy
    dtype: string
  - name: coordinates_obscured
    dtype: bool
  - name: positioning_method
    dtype: string
  - name: positioning_device
    dtype: string
  - name: species_guess
    dtype: string
  - name: scientific_name
    dtype: string
  - name: common_name
    dtype: string
  - name: iconic_taxon_name
    dtype: string
  - name: taxon_id
    dtype: int64
  - name: sound_format
    dtype: string
  splits:
  - name: train
    num_bytes: 48900983
    num_examples: 68824
  - name: val
    num_bytes: 5316841
    num_examples: 7501
  download_size: 21473814
  dataset_size: 54217824
- config_name: clip_tri_modal
  features:
  - name: id
    dtype: int64
  - name: sat_key
    dtype: string
  - name: sat_path
    dtype: string
  - name: sat_bounds
    struct:
    - name: max_lat
      dtype: float64
    - name: max_lon
      dtype: float64
    - name: min_lat
      dtype: float64
    - name: min_lon
      dtype: float64
  - name: img_ids
    sequence: int64
  - name: target_positions
    sequence:
      sequence: int64
  splits:
  - name: train
    num_bytes: 114789736
    num_examples: 379962
  - name: val
    num_bytes: 9472004
    num_examples: 57752
  download_size: 62896043
  dataset_size: 124261740
- config_name: search_quad_modal
  features:
  - name: id
    dtype: int64
  - name: sat_key
    dtype: string
  - name: sat_path
    dtype: string
  - name: taxonomy
    dtype: string
  - name: count
    dtype: int64
  - name: spread
    dtype: float64
  - name: sat_bounds
    struct:
    - name: max_lat
      dtype: float64
    - name: max_lon
      dtype: float64
    - name: min_lat
      dtype: float64
    - name: min_lon
      dtype: float64
  - name: img_ids
    sequence: int64
  - name: target_positions
    sequence:
      sequence: int64
  - name: num_landmarks
    dtype: int64
  - name: sound_ids
    sequence: int64
  splits:
  - name: train
    num_bytes: 5235058
    num_examples: 10805
  - name: val_in_domain
    num_bytes: 155806
    num_examples: 459
  - name: val_out_domain
    num_bytes: 134064
    num_examples: 391
  download_size: 2844508
  dataset_size: 5524928
- config_name: search_tri_modal
  features:
  - name: id
    dtype: int64
  - name: sat_key
    dtype: string
  - name: sat_path
    dtype: string
  - name: taxonomy
    dtype: string
  - name: count
    dtype: int64
  - name: spread
    dtype: float64
  - name: sat_bounds
    struct:
    - name: max_lat
      dtype: float64
    - name: max_lon
      dtype: float64
    - name: min_lat
      dtype: float64
    - name: min_lon
      dtype: float64
  - name: img_ids
    sequence: int64
  - name: target_positions
    sequence:
      sequence: int64
  - name: num_landmarks
    dtype: int64
  splits:
  - name: train
    num_bytes: 27066170
    num_examples: 80535
  - name: val_in_domain
    num_bytes: 1208012
    num_examples: 4000
  - name: val_out_domain
    num_bytes: 1319681
    num_examples: 4000
  download_size: 13604442
  dataset_size: 29593863
configs:
- config_name: clip_quad_modal
  data_files:
  - split: train
    path: clip_quad_modal/train-*
  - split: val
    path: clip_quad_modal/val-*
- config_name: clip_tri_modal
  data_files:
  - split: train
    path: clip_tri_modal/train-*
  - split: val
    path: clip_tri_modal/val-*
- config_name: search_quad_modal
  data_files:
  - split: train
    path: search_quad_modal/train-*
  - split: val_in_domain
    path: search_quad_modal/val_in_domain-*
  - split: val_out_domain
    path: search_quad_modal/val_out_domain-*
- config_name: search_tri_modal
  data_files:
  - split: train
    path: search_tri_modal/train-*
  - split: val_in_domain
    path: search_tri_modal/val_in_domain-*
  - split: val_out_domain
    path: search_tri_modal/val_out_domain-*
---

# Model Card for Search-TTA-Dataset
Dataset to finetune [search_tta_sat](https://huggingface.co/derektan95/search-tta-sat) and [search_tta_sound](https://huggingface.co/derektan95/search-tta-sound) CLIP models.

For more information on usage, please refer to the [Search-TTA-VLN Github repository here](https://github.com/marmotlab/Search-TTA-VLN).

<img src="avs_bench_examples.jpg" width="100%"/>

## Citation
```
@inproceedings{tan2025searchtta,
  title     = {Search-TTA: A Multi-Modal Test-Time Adaptation Framework for Visual Search in the Wild},
  author    = {Tan, Derek Ming Siang and Shailesh, Shailesh and Liu, Boyang and Raj, Alok and Ang, Qi Xuan and Dai, Weiheng and Duhan, Tanishq and Chiun, Jimmy and Cao, Yuhong and Shkurti, Florian and Sartoretti, Guillaume Adrien},
  booktitle = {Proceedings of The 9th Conference on Robot Learning},
  pages     = {2093--2120},
  year      = {2025},
  volume    = {305},
  publisher = {PMLR}
}

@inproceedings{sastry2025taxabind,
  title     = {Taxabind: A unified embedding space for ecological applications},
  author    = {Sastry, Srikumar and Khanal, Subash and Dhakal, Aayush and Ahmad, Adeel and Jacobs, Nathan},
  booktitle = {2025 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
  pages     = {1765--1774},
  year      = {2025},
  url       = {https://arxiv.org/abs/2505.11350}
}
```