A system for identifying different apple varieties based on the YOLOV8X neural network
| Authors: Mikheev D.A., Kitaev D.N. |  |
| Published in issue: #3(92)/2024 | |
| DOI: | |
Category: Informatics, Computer Engineering and Control | Chapter: Information Technology. Computer techologies. Theory of computers and systems |
|
Keywords: computer vision, object identification, neural networks, convolutional neural network, single-stage detector, YOLOv8x, weighed products, apple varieties, retail |
|
| Published: 28.07.2024 | |
The paper presents results of selecting a neural network model to solve the problem of identifying the weighed products. It considers a family of the single-stage models of the YOLOv8 convolutional neural networks and assesses at the initial stage performance of the largest YOLOv8x model on the frames with images of fruits and vegetables in a grocery store. Data were collected and prepared for the assessed network learning to recognize five apple varieties: Golden Delicious, Granny Smith, Gala, Honey Crisp and Red Chief. The obtained data was introduced to learn the YOLOv8x model using the transfer learning; results of the learned model operation were analyzed.
References
[1] Andrianov N.A., Dementyev V.E., Tashlinsky A.G. Object detection in images: from Bayes and Neyman-Pearson criteria to detectors based on neural networks EfficientDet. Computer Optics, 2022, Vol. 46, No. 1, pp. 139–159. https://doi.org/10.18287/2412-6179-CO-922
[2] Albumentations, “Computer vision tool that boosts the performance of deep convolutional neural networks”. Available at: https://albumentations.ai/ (accessed April 15, 2024).
[3] Computer Hope, Dictionary, Definitions, “FLOPS”. Available at: https://www.computerhope.com/jargon/f/flops.htm (accessed April 15, 2024).
[4] GitHub, ultralytics repository, YOLOv8. URL: https://github.com/ultralytics/ultralytics (accessed April 15, 2024).
[5] Hui J. mAP (mean Average Precision) for object detection. Available at: https://jonathan-hui.medium.com/map-meanaverage-precision-for-object-detection-45c121a31173 (accessed April 15, 2024).
[6] Alzubaidi L., Zhang J., Humaidi A.J. Review of deep learning: concepts, CNN architectures, challenges, applications, future directions. Journal of Big Data, 2021, art. 53. https://doi.org/10.1186/s40537-021-00444-8
[7] Open Neural Network Exchange. Available at: https://onnx.ai/ (accessed April 15, 2024).
[8] Powers D. Evaluation: From precision, recall and f-measure to ROC, Informedness, markedness & correlation. Journal of Machine Learning Technologies, 2011, vol. 2 (1), pp. 37–63.
[9] Redmon J., Divvala S., Girshick R., Farhadi A. You only look once: Unified, real-time object detection. Proc IEEE Conf on Computer Vision and Pattern Recognition (CVPR), 2016, vol. 1, pp. 779–788. https://doi.org/10.48550/arXiv.1506.02640
[10] Ockay M., Harakal M., Liska M. Compute Unified Device Architecture (CUDA) GPU programming model and possible integration to the parallel environment. Science & Military, no. 2, vol. 3, 2008, pp. 64–68.
[11] Roboflow, Jacob Solawetz, Francesco “What is YOLOv8? The Ultimate Guide.” Available at: https://blog.roboflow.com/whats-new-in-yolov8/ (accessed April 15, 2024).
[12] Roboflow, “Quickly Label Training Data and Export to Any Format.” Available at: https://roboflow.com/annotate (accessed April 15, 2024).