FishNet: Deep Neural Networks for Low-Cost Fish Stock Estimation

Moseli Mots'oehli,Anton Nikolaev,Wawan B. IGede,John Lynham,Peter J. Mous,Peter Sadowski

2024-06-28

Abstract:Fish stock assessment often involves manual fish counting by taxonomy specialists, which is both time-consuming and costly. We propose FishNet, an automated computer vision system for both taxonomic classification and fish size estimation from images captured with a low-cost digital camera. The system first performs object detection and segmentation using a Mask R-CNN to identify individual fish from images containing multiple fish, possibly consisting of different species. Then each fish species is classified and the length is predicted using separate machine learning models. To develop the model, we use a dataset of 300,000 hand-labeled images containing 1.2M fish of 163 different species and ranging in length from 10cm to 250cm, with additional annotations and quality control methods used to curate high-quality training data. On held-out test data sets, our system achieves a 92% intersection over union on the fish segmentation task, a 89% top-1 classification accuracy on single fish species classification, and a 2.3cm mean absolute error on the fish length estimation task.

Computer Vision and Pattern Recognition,General Economics

What problem does this paper attempt to address?

The paper aims to address the high cost and time-consuming issues in fishery resource assessment, particularly in the context of developing countries. Specifically, the paper proposes the FishNet system, an automated method based on computer vision for fish species classification and size estimation. Traditional fish resource assessment methods rely on professional taxonomists to manually count and classify fish, which is both time-consuming and costly. The FishNet system significantly reduces assessment costs by using photos taken with low-cost digital cameras, combined with deep learning techniques for fish identification, segmentation, and size estimation. Additionally, the system overcomes some limitations of traditional methods, such as the need for specialized equipment or operation in underwater environments, making it a more practical and cost-effective solution. The research results show that the FishNet system achieved a 92% intersection over union in the fish segmentation task, an 89% accuracy in the single fish classification task, and an average absolute error of 2.3 centimeters in the fish length estimation task. These results indicate that the FishNet system can effectively improve the efficiency and accuracy of fishery resource assessment, especially suitable for resource-limited developing countries.

FishNet: Deep Neural Networks for Low-Cost Fish Stock Estimation

Multi-class Fish Stock Statistics Technology Based on Object Classification and Tracking Algorithm

An Automatic Detection and Statistical Method for Underwater Fish Based on Foreground Region Convolution Network (FR-CNN)

Accurate Fish Detection under Marine Background Noise Based on the Retinex Enhancement Algorithm and CNN

Fish Keypoints Detection for Ecology Monitoring Based on Underwater Visual Intelligence

Image-based, unsupervised estimation of fish size from commercial landings using deep learning

FishNet: A Unified Embedding for Salmon Recognition

Fish Monitoring from Low-Contrast Underwater Images

Multi-classification deep neural networks for identification of fish species using camera captured images

Assessing fish abundance from underwater video using deep neural networks

Computer vision in aquaculture: a case study of juvenile fish counting

Deep but Lightweight Neural Networks for Fish Detection

ConvFishNet: an Efficient Backbone for Fish Classification from Composited Underwater Images

An Automatic Recognition Method for Fish Species and Length Using an Underwater Stereo Vision System

Deep Convolutional Neural Networks for Fish Weight Prediction from Images

Automatic Weight Estimation of Harvested Fish from Images

A deep neural network for multi-species fish detection using multiple acoustic cameras

Automatic measurement of fish from images using convolutional neural networks

Crosspooled FishNet: transfer learning based fish species classification model

The Fishnet Open Images Database: A Dataset for Fish Detection and Fine-Grained Categorization in Fisheries

Enhanced deep learning models for automatic fish species identification in underwater imagery