AI Shrimp Counting Application
An aquaculture counting application using YOLO and MobileNet to explore image-based shrimp counting.
The counting problem
Counting shrimp manually is repetitive work, particularly when individuals overlap or are difficult to distinguish in an image. This project applies computer vision to shrimp counting: a concrete agricultural use case where the quality of the input image matters as much as the choice of model.
The objective is to turn visual observations into a useful count that an operator can assess. A count should be considered alongside the capture conditions and the possibility of missed or duplicate detections.
YOLO and MobileNet
Risenine’s project uses YOLO and MobileNet. YOLO is an object detection model family that can locate objects within an image. MobileNet is a lightweight neural network architecture family suited to applications with constrained computing resources.
Using these technologies in a counting application brings together object detection and efficient visual processing. The exact roles of the models, training dataset, and deployment hardware depend on the implementation; no particular model version or device specification is claimed here.
What makes shrimp counting difficult?
Lighting, reflections, water clarity, camera distance, animal size, and overlapping shrimp can change the appearance of the same scene. A useful evaluation should include these conditions rather than rely only on clear, isolated images.
For a counting system, detection accuracy alone is insufficient. Count error against a manually checked reference, repeated measurements, and processing time are relevant ways to assess suitability for an aquaculture workflow.
Business relevance and project scope
This project demonstrates Risenine’s focus on applying AI to a specific operational task. It is relevant to teams exploring visual counting, object detection, or computer vision applications in agriculture and aquaculture.
Accuracy percentages, production deployment status, and commercial time savings are not published for this project. A project discussion can establish the available evidence, data requirements, capture setup, and validation needed for your use case.
Discuss your AI project.
Share the problem, available data, and workflow you want to improve.
Discuss your project