Models and Datasets utilized by GhostVision, a utility that performs near-real time detection of derelict (ghost) crab pots with side-scan sonar.

Publication

In review at Journal of Marine Science and Engineering

Ghost Pot Side‑Scan Sonar Detection Dataset

This dataset contains manually annotated side‑scan sonar (SSS) imagery collected across Delaware’s Inland Bays and Delaware Bay to support research on automated detection of derelict crab pots (“ghost pots”). It is designed for training and evaluating object‑detection models in turbid, shallow‑water environments where visual surveys are limited and acoustic mapping is essential.

The dataset accompanies the GhostVision project, an open‑source pipeline for near–real‑time detection and mapping of derelict fishing gear.

More details available on HuggingFace.

RF-DETR model for side-scan sonar ghost pot detection

This repository contains a fine-tuned RF-DETR object-detection model for identifying derelict crab pots in side-scan sonar imagery. The model was trained on the PINGEcosystem/sss-crab-pot-detection-ds dataset, which contains annotated sonar imagery collected in Delaware's Inland Bays and Delaware Bay.

The model is part of the GhostVision effort to support scalable detection and mapping of derelict fishing gear from acoustic imagery.

More details available on HuggingFace.

YOLO12 model for side-scan sonar ghost pot detection

This repository contains a fine-tuned YOLO-based object-detection model for identifying derelict crab pots in side-scan sonar imagery. The model was trained on the PINGEcosystem/sss-crab-pot-detection-ds dataset, which contains annotated sonar imagery collected in Delaware's Inland Bays and Delaware Bay.

The model is part of the GhostVision effort to support scalable detection and mapping of derelict fishing gear from acoustic imagery.

More details available on HuggingFace.

YOLO26 model for side-scan sonar ghost pot detection

This repository contains a fine-tuned YOLO-based object-detection model for identifying derelict crab pots in side-scan sonar imagery. The model was trained on the PINGEcosystem/sss-crab-pot-detection-ds dataset, which contains annotated sonar imagery collected in Delaware's Inland Bays and Delaware Bay.

The model is part of the GhostVision effort to support scalable detection and mapping of derelict fishing gear from acoustic imagery.

More details available on HuggingFace.