Related Experiment Video
Updated: Jul 4, 2025
13:19
Deep Neural Networks for Image-Based Dietary Assessment
Published on: March 13, 2021
9.2K
Computer vision based deep learning approach for toxic and harmful substances detection in fruits
Abdus Sattar1,2, Md Asif Mahmud Ridoy2, Aloke Kumar Saha3
1Centre for Higher Studies and Research, Bangladesh University of Professionals, Dhaka, Bangladesh.
Heliyon
|February 8, 2024
Summary
Detecting toxic formaldehyde (CH₂O) in fruits is challenging. Deep learning models, especially the novel DurbeenNet, show high accuracy in identifying chemically preserved produce, ensuring food safety.
Area of Science:
- Food Science
- Computer Science
- Chemistry
Background:
- Formaldehyde (CH₂O) is illicitly used to preserve perishable fruits in Bangladesh.
- Chemical detection is difficult due to its undetectable nature.
- There is a need for reliable detection methods.
Purpose of the Study:
- To develop and evaluate deep learning models for detecting formaldehyde in fruits.
- To compare the performance of pre-trained models with a novel model, DurbeenNet.
Main Methods:
- Applied deep learning techniques to detect toxic substances in four fruit types.
- Utilized data augmentation to expand the dataset.
- Assessed GoogleNet, VGG-16, DenseNet, ResNet50, and the proposed DurbeenNet model.
Main Results:
- ResNet50 achieved 91.66% accuracy, DenseNet 90.37%, VGG-16 87.44%, and GoogleNet 85.53%.
- The novel DurbeenNet model demonstrated superior performance with 96.71% accuracy.
- DurbeenNet significantly outperformed established deep learning architectures.
Conclusions:
- Deep learning, particularly DurbeenNet, offers a robust solution for detecting formaldehyde in fruits.
- The developed model can enhance food safety by identifying artificially preserved produce.
- This technology addresses a critical challenge in food quality control.
