Research map / Detection and segmentation / Tracking
The Caltech Fish Counting Dataset: A Benchmark for Multiple-Object Tracking and Counting
Justin Kay, Peter Kulits, Suzanne Stathatos, Siqi Deng, Erik Young, Sara Beery, Grant Van Horn, Pietro Perona
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 background-subtracted and frame-difference input channels (Baseline++) (ablation) | YOLOv5 + SORT · Caltech Fish Counting KL (val) | nMAE (%): 4.9% -> 3.3% | Improved |
| add background-subtracted and frame-difference input channels (Baseline++) (ablation) | YOLOv5 + SORT · Caltech Fish Counting KL (val) | AP50: +1.6 AP50 | Improved |
| add background-subtracted and frame-difference input channels (Baseline++) (ablation) | YOLOv5 + SORT · Caltech Fish Counting KC | nMAE (%): 53.0% -> 12.8% | Improved |
| add background-subtracted and frame-difference input channels (Baseline++) (ablation) | YOLOv5 + SORT · Caltech Fish Counting EL | nMAE (%): 32.3% -> 21.3% | Improved |
| add background-subtracted and frame-difference input channels (Baseline++) (ablation) | YOLOv5 + SORT · Caltech Fish Counting KR | nMAE (%): 11.8% -> 3.7% | Improved |
| add background-subtracted and frame-difference input channels (Baseline++) (ablation) | YOLOv5 + SORT · Caltech Fish Counting EL | MOTA: -376.7 -> -54.5 | Improved |
| give the tracker ground-truth detections (Perfect Detector) (ablation) | YOLOv5 + SORT · Caltech Fish Counting | counting error | Improved |
| fit tracker hyperparameters to each test set (Perfect Tracker Generalization) (ablation) | YOLOv5 + SORT · Caltech Fish Counting | counting error | Improved |
| add appearance re-identification features for association (ResNet-50 triplet) evaluated on this dataset | SORT · Caltech Fish Counting (Val) | mAP: 80.7 -> 19.2 | Hurt |
| choose YOLOv5m over Faster R-CNN + Resnet101 detector | detector · Caltech Fish Counting validation (detector choice) | AP50: 65.9 -> 66.4 | Improved |
Closest papers by meaning
- When Trackers Date Fish: A Benchmark and Framework for Underwater Multiple Fish Tracking (AAAI 2026 · similarity 0.90)
- CFC26: Building Evaluations for Deployment in Sonar-Based Fish Counting (NeurIPS 2026 · similarity 0.89)
- CountGD: Multi-Modal Open-World Counting (NeurIPS 2024 · similarity 0.85)
- WebUOT-1M: Advancing Deep Underwater Object Tracking with A Million-Scale Benchmark (NeurIPS 2024 · similarity 0.85)
- OVT-B: A New Large-Scale Benchmark for Open-Vocabulary Multi-Object Tracking (NeurIPS 2024 · similarity 0.85)
- DAVE - A Detect-and-Verify Paradigm for Low-Shot Counting (CVPR 2024 · 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.
