Automated Pork Cut and Freshness Detection Using Computer Vision
Author Information
Author(s): Song Shihao, Guo Qiqi, Duan Xiaosa, Shi Xiaojing, Liu Zhenyu
Primary Institution: Shanxi Agricultural University
Hypothesis
Can computer vision technology improve the accuracy and efficiency of pork quality assessment?
Conclusion
The MobileNetV3_Small model achieved an accuracy of 99.71% in recognizing and classifying different cuts and freshness levels of pork.
Supporting Evidence
- The MobileNetV3_Small model outperformed other models with an accuracy of 98.59%.
- Data augmentation significantly improved model performance.
- Statistical analysis showed significant differences in performance between models.
Takeaway
This study shows that a computer can help tell if pork is fresh and what part it comes from, making it easier to check meat quality.
Methodology
The study used high-resolution cameras to capture images of pork cuts and applied various convolutional neural network models for analysis.
Potential Biases
The model may struggle with subtle features related to pork freshness due to lighting inconsistencies.
Limitations
Lighting variations and camera quality can affect the accuracy of the model.
Participant Demographics
Images were collected from Jinfen white pigs, a locally bred lean-type pig.
Statistical Information
P-Value
0.0037
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website