Large Scale Fish Images Classification and Localization using Transfer Learning and Localization Aware CNN Architecture

Abstract

Building an automatic fish recognition and detection system for largescale fish classes is helpful for marine researchers and marine scientists because there are large numbers of fish species. However, it is quite difficult to build such systems owing to the lack of data imbalance problems and large number of classes. To solve these issues, we propose a transfer learning-based technique in which we use Efficient-Net, which is pre-trained on ImageNet dataset and fine-tuned on QuT Fish Database, which is a large scale dataset. Furthermore, prior to the activation layer, we use Global Average Pooling (GAP) instead of dense layer with the aim of averaging the results of predictions along with having more information compared to the dense layer. To check the validity of our model, we validate our model on the validation set which achieves satisfactory results. Also, for the localization task, we propose an architecture that consists of localization aware block, which captures localization information for better prediction and residual connections to handle the over-fitting problem. Actually, the residual connections help the layer to combine missing information with the relevant one. In addition, we use class weights and Focal Loss (FL) to handle class imbalance problems along with reducing false predictions. Actually, class weights assign less weights to classes having fewer instances and large weights to classes having more number of instances. During the localization, the qualitative assessment shows that we achieve 57% Mean Intersection Over Union (IoU) on testing data, and the classification results show 75% precision, 70% recall, 78% accuracy and 74% F1-Score for 468 fish species.

Aquasafe: A Remote Sensing, Web-Based Platform for the Support of Precision Fish Farming

“…The integration of data derived from UAVs, which have already been widely used in land and sea applications [59][60][61][62], can provide a wealth of data for managing fish farms. By utilizing cameras, sensors, and other instruments, drones can capture high-resolution images and data on various important parameters for fish farm management such as water quality, fish health and behavior, farm infrastructure, and environmental conditions [63][64][65][66]. Finally, incorporating sea surface current data from remote sensing and numerical models can provide valuable information on the direction, speed, and variability of the currents.…”

Classification Of Some Edible Submarine Organisms in The Mediterranean Region by Species Using Deep Learning Methods

“…A transfer learning based pre-trained model on ImageNet was used to improve the classification performance. In the study, a CNN-based architecture was studied [15]. For the categorization of fish images, Aziz et al suggested a Deep Learning Artificial Neural Network (DLANN) model with a cutting-edge optimization method.…”

Stacking Ensemble Approach for Churn Prediction: Integrating CNN and Machine Learning Models with CatBoost Meta-Learner

“…The empirical results In recent years, the rise of deep learning has revolutionized the field of churn prediction. Ahmed et al [14] proposed a model, called TL-DeepE, for churn prediction. This model employed the principle of Transfer Learning and Ensemble-based Meta-Classification.…”