Automated Steak Marbling Grading Solution

Azati developed a mobile application powered by a CNN model to assess steak marbling levels based on photos, ensuring objective quality grading and faster selection processes.

Discuss an idea

All Technologies Used

Python
Python
PyTorch
PyTorch
PyTorch Lightning
PyTorch Lightning
Pandas
Pandas
OpenCV
OpenCV
React Native
React Native
Docker
Docker

Motivation

The client, a major player in the U.S. Agricultural sector, needed a tool to standardize meat quality evaluation by grading steak marbling. Azati developed a solution to streamline this process by training a CNN model to detect subtle marbling differences and integrating it into a user-friendly mobile application, enhancing consistency and efficiency in steak grading.

Main Challenges

Challenge 1
Ensuring efficient communication with a busy client

The client’s frequent business trips delayed responses and project approvals. To mitigate this, Azati transitioned to written communication supported by prototypes and clear examples, allowing faster decision-making and project alignment.

Challenge 2
Eliminating lighting glare in steak analysis

Steaks stored under bright refrigeration lights caused glare that AI models misinterpreted as fat, leading to overestimated marbling grades. To solve this, the client developed a custom ‘hood’ to block external lighting and ensure accurate image analysis.

Key Features

  • CNN-Powered Image Analysis: Leverages Convolutional Neural Networks to evaluate steak marbling levels from photos.
  • User-Friendly Mobile Interface: Provides an intuitive platform for users to capture images and receive instant quality grading.
  • Model Update Flexibility: Allows seamless AI model updates without the need to rebuild the app.
  • Iteration-Based Functionality Enhancement: Refined through multiple iterations based on real-world user feedback.

Our Approach

Understanding Client Needs
Azati conducted a discovery call with the client to understand requirements and explore current market solutions for marbling assessment, creating a tailored approach for the project.
Training a CNN Model for Marbling Detection
Azati trained a CNN using a client-provided dataset, ensuring the model could accurately detect and classify marbling levels across various steak images.
Mobile App Development and Iterative Refinement
The mobile application was designed and developed as an MVP, allowing users to capture photos and assess marbling levels. User feedback led to three iterations, further enhancing functionality and user experience.
Enabling Seamless Model Updates
The app was built to allow easy updates of the AI model without requiring a full rebuild, ensuring scalability and maintainability.

Project Impact

Azati’s solution delivered measurable improvements in steak grading efficiency and accuracy. The trained CNN model ensures consistent quality assessment, while the flexible app design supports long-term scalability and usability. The client can now provide reliable, objective grading standards with faster processing times, enhancing operational efficiency in meat evaluation.

Ready To Get Started

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.