Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper details the development of a multi-stage detection system for the FathomNetCLEF 2026 competition, which faced challenges of sparse training labels and out-of-distribution test data in marine species detection. The authors utilized a frozen Megalodon YOLOv8x detector for proposal generation, combined with a LoRA-finetuned DINOv3 ViT-H classifier and a novel reranking mechanism to enhance prediction accuracy. Their approach achieved a commendable 12th place out of 102 teams, highlighting the importance of effective validation strategies over traditional metrics in model selection.
Relying on proxy datasets for validation rather than train-derived metrics proved crucial in achieving competitive performance in marine species detection.
The FathomNetCLEF 2026 competition combines underwater object detection and fine-grained marine species classification under a positive-unlabeled evaluation setting. The provided training labels are sparse, while the hidden test set is out-of-distribution relative to the training imagery, creating both annotation incompleteness and source-shift challenges. We describe DS@GT ARC's multi-stage system developed for this setting while keeping model training restricted to the data provided by the competition. The final private-leaderboard model uses a frozen Megalodon YOLOv8x detector as a class-agnostic proposal generator, combines global and tiled inference with tile-edge filtering, classifies expanded proposal crops with a LoRA-finetuned DINOv3 ViT-H classifier, and ranks predictions using weighted geometric fusion of detector and classifier confidence. This system placed 12th out of 102 teams. A closely related variant added a locally trained TTN-inspired validity head as a light reranking signal, improving public-leaderboard and proxy-evaluation performance but slightly reducing private-leaderboard performance. Across experiments, the strongest lesson was that train-derived validation and detector-only metrics were not reliable enough for model selection. Instead, we used proxy datasets only for validation and comparison, and combined those signals with leaderboard feedback and targeted ablations. These experiments showed that reserving proposal recall, avoiding over-aggressive filtering, and improving downstream ranking were more effective than fine-tuning the detector or directly training on noisy pseudo-labels. Code: https://github.com/dsgt-arc/fathomnetclef-2026.