Scholarlink

This research paper presents a novel attempt for the classification of fish freshness using a hybrid Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) model. The study primarily focuses on distinguishing between fresh and non-fresh fish based on images of their eyes and gills. The dataset utilized in this study comprises labeled images, which were processed and augmented using the ImageDataGenerator from the TensorFlow library to improve model robustness. The architecture of the proposed model integrates the spatial feature extraction capability of CNNs with the temporal sequence learning ability of LSTMs. This hybrid model is constructed with three convolutional layers followed by max-pooling layers, a flattening layer, and a dense layer before reshaping the output for the LSTM layer. The model was trained over 20 epochs with a batch size of 32, and Adam optimizer was employed for compiling the model with a learning rate of 0.001. Experimental results demonstrate that the hybrid CNN-LSTM model achieves an overall accuracy of 86% on the test dataset. The model’s performance was assessed using confusion matrices, classification reports, and precision-recall-F1 score metrics, revealing its effectiveness in accurately classifying fish freshness. Additionally, Local Interpretable Model-agnostic Explanations (LIME) were employed to provide insights into the model's decision-making process, enhancing the interpretability of the classification results. This study underscores the potential of combining CNNs and LSTMs for image-based classification tasks in the domain of fish freshness assessment, providing a robust tool for ensuring food quality and safety.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Fish Freshness Detection Via Hybrid CNN-LSTM: An Interpretable Deep Learning Model

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Sitanath Biswas,
  • Chirag Nahata,
  • Snigdha Ghosh,
  • Shubhashree Sahoo,
  • Dipanjana Biswas

摘要

This research paper presents a novel attempt for the classification of fish freshness using a hybrid Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) model. The study primarily focuses on distinguishing between fresh and non-fresh fish based on images of their eyes and gills. The dataset utilized in this study comprises labeled images, which were processed and augmented using the ImageDataGenerator from the TensorFlow library to improve model robustness. The architecture of the proposed model integrates the spatial feature extraction capability of CNNs with the temporal sequence learning ability of LSTMs. This hybrid model is constructed with three convolutional layers followed by max-pooling layers, a flattening layer, and a dense layer before reshaping the output for the LSTM layer. The model was trained over 20 epochs with a batch size of 32, and Adam optimizer was employed for compiling the model with a learning rate of 0.001. Experimental results demonstrate that the hybrid CNN-LSTM model achieves an overall accuracy of 86% on the test dataset. The model’s performance was assessed using confusion matrices, classification reports, and precision-recall-F1 score metrics, revealing its effectiveness in accurately classifying fish freshness. Additionally, Local Interpretable Model-agnostic Explanations (LIME) were employed to provide insights into the model's decision-making process, enhancing the interpretability of the classification results. This study underscores the potential of combining CNNs and LSTMs for image-based classification tasks in the domain of fish freshness assessment, providing a robust tool for ensuring food quality and safety.