2022 International Congress on Human-Computer Interaction, Optimization and Robotic Applications (HORA) · 2022 · 16 citations · 20 references
ComputingComputer ScienceComputer VisionFruit ScienceMachine LearningDeep LearningFruit QualityImage ClassificationObject RecognitionPattern RecognitionRepresentation LearningAccuracyBinary ClassificationConvolutional Neural NetworksDeep Learning-based ClassificationGrid SearchHyperparameter TuningK-fold Cross-validationK-fold Validation
Automatic freshness classification of fruits and veg-etables is an active area of research in the modern era. In order to maintain and preserve the grade quality of the fruits, the need to speed up processing is in a great urge. The fruit Hog Plum is noted to have high demand in Indian subcontinents. In this work we have utilized deep learning based techniques for performing freshness classification on hog Plum fruit. In this work, four convolutional neural network based models viz. ResNet18, MobileNetV2, MobileNetV3-Small, MobileNetV3-Large is pre-trained on the ImageNet dataset which were fine-tuned to perform binary classification on Hog Plum dataset cre-ated by us. The dataset was augmented for training purpose and hyperparameter tuning was performed using grid search and k-fold cross validation. The results of the model are then evaluated and compared in terms of several metrics like accuracy, precision, recall, Fl-Score, ROC-AUC Score and balanced accuracy score. Our results demonstrate that deep learning based techniques are effective for performing freshness classification of Hog Plum fruit. The models trained by us are lightweight in nature, hence are suitable to run on low-end and mobile devices increasing the applicability and adaptability of our proposed method.
20
Deep Residual Learning for Image Recognition
Kaiming He, Xiangyu Zhang, Shaoqing Ren et al. · 2016 · 222.1K citations · Full text
ImageNet: A large-scale hierarchical image database
Jia Deng, Wei Dong, Richard Socher et al. · 2009 IEEE Conference on Computer Vision and Pattern Recognition · 2009 · 61.8K citations
MobileNetV2: Inverted Residuals and Linear Bottlenecks
Mark Sandler, Andrew Howard, Menglong Zhu et al. · 2018 · 25.4K citations
AutoAugment: Learning Augmentation Strategies From Data
Ekin D. Cubuk, Barret Zoph, Dandelion Mané et al. · 2019 · 2.7K citations