Research map / Detection and segmentation / Open-set and few-shot detection
CHIRP dataset: towards long-term, individual-level, behavioral monitoring of bird populations in the wild
Alex Hoi Hang Chan, Neha Singhal, Onur Kocahan, Andrea Meltzer, Saverio Lubrano, Miyako H. Warrington, Michael Griesser, Fumihiro Kano, Hemal Naik
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 |
|---|---|---|---|
| constrain candidate IDs to within-territory list for CORVID (ablation) | CORVID · CHIRP re-ID closed set | top-1 accuracy: 0.05 -> 0.66 | Improved |
| use CORVID color-ring matching instead of pre-trained MegaDescriptor (within territory) | re-ID · CHIRP closed set | top-1 accuracy: 0.28 -> 0.66 | Improved |
| use CORVID instead of MegaDescriptor when gallery includes all individuals | re-ID · CHIRP closed set (all) | top-1 accuracy: 0.10 -> 0.05 | Hurt |
| fine-tune MegaDescriptor on CHIRP | MegaDescriptor · CHIRP closed set within territory | top-1 accuracy: 0.28 -> 0.27 | No clear effect |
| fine-tune MegaDescriptor on CHIRP | MegaDescriptor · CHIRP disjointed set within territory | top-1 accuracy: 0.31 -> 0.41 | Improved |
| use CORVID re-ID in pipeline instead of MegaDescriptor | tracking+action pipeline · CHIRP application-specific benchmark | proportion correct frame assignments: 0.617 -> 0.647 | Improved |
| use CORVID instead of random ID assignment | pipeline · CHIRP application-specific benchmark | feeding-rate mean absolute error: 9.349 -> 9.000 | No clear effect |
| use fine-tuned MegaDescriptor instead of random ID assignment | pipeline · CHIRP application-specific benchmark | feeding-rate mean absolute error: 9.349 -> 13.14 | Hurt |
| use C3D instead of SlowFast for action recognition | action recognition · CHIRP action recognition | top-1 accuracy: 0.678 -> 0.715 | Improved |
| use ViTPose-large instead of ResNet-50 | keypoint estimation · CHIRP 2D keypoints | PCK@5: 0.832 -> 0.915 | Improved |
Closest papers by meaning
- BirdCollect: A Comprehensive Benchmark for Analyzing Dense Bird Flock Attributes (AAAI 2024 · similarity 0.89)
- The PanAf-FGBG Dataset: Understanding the Impact of Backgrounds in Wildlife Behaviour Recognition (CVPR 2025 · similarity 0.89)
- MammAlps: A Multi-view Video Behavior Monitoring Dataset of Wild Mammals in the Swiss Alps (CVPR 2025 · similarity 0.89)
- BuckTales: A multi-UAV dataset for multi-object tracking and re-identification of wild antelopes (NeurIPS 2024 · similarity 0.89)
- Generating the Wild: Individual-Consistent Image-to-Video Generation for Wildlife (NeurIPS 2026 · similarity 0.88)
- RealBirdID: Benchmarking Bird Species Identification in the Era of MLLMs (CVPR 2026 · similarity 0.86)
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.
