About

Starting Semester: Fall 2026
Assigned: No
Location: Atlanta

KEH Camera

Client Profile

KEH Camera is one of the largest global buyers and sellers of pre-owned photography and videography equipment. The company operates a vertically integrated recommerce platform that sources used cameras, lenses, and accessories from individual sellers and resells them to photographers, videographers, and content creators worldwide.

The company specializes in the refurbishment, grading, and resale of professional and consumer imaging equipment across major brands such as Canon, Nikon, Sony, Fujifilm, Leica, and others. All products are inspected and graded through a standardized evaluation process conducted by trained technicians.

KEH operates primarily through its direct-to-consumer e-commerce platform (keh.com) and supports customers in over 160 countries. The company also purchases used gear directly from consumers, professional photographers, retailers, and institutional sellers.

Project Description

The resale market for professional camera equipment relies heavily on accurate condition grading to determine product pricing, resale value, and customer expectations. Currently, grading processes for used camera bodies and lenses are primarily performed by trained technicians through manual visual inspection. While experienced graders can achieve consistent results, the process can be time-intensive and subject to human variability when evaluating cosmetic wear and physical condition.

This project will explore the development of an AI-assisted imaging system capable of supporting or partially automating the grading of used photography equipment. The goal is to design a system that captures standardized images of camera bodies and lenses and applies computer vision and machine learning techniques to detect cosmetic wear, defects, and condition indicators relevant to resale grading.

The proposed system would combine controlled imaging hardware and machine learning models to evaluate equipment condition across multiple dimensions, such as:
- Surface wear, scratches, and cosmetic damage
- Lens glass condition and visible defects
- Physical deformation or impact damage
- Missing components or accessories
- Overall visual condition consistent with resale grading standards

Students will investigate approaches for building a structured imaging pipeline, including lighting, positioning, and capture angles that allow consistent image collection. The project may also explore computer vision methods such as object detection, defect detection, image segmentation, and classification models to identify and quantify condition indicators.

In addition to model development, the team will analyze how an automated or semi-automated grading system could integrate into existing refurbishment and intake workflows. This includes evaluating accuracy relative to human graders, throughput improvements, and potential impacts on operational efficiency.

Key deliverables for the project may include:
- Design of a standardized imaging setup for used equipment
- A prototype computer vision model for detecting cosmetic condition indicators
- A framework for mapping visual signals to resale grading categories
- Performance evaluation comparing AI-assisted grading with human inspection
- Recommendations for operational deployment within a resale environment

The ultimate objective is to determine whether computer vision and AI can improve grading consistency, increase processing speed, and support scalable operations in the recommerce of high-value electronics.

Skills

Students with experience or coursework in the following areas are preferred:
- Python programming (NumPy, Pandas, OpenCV, PyTorch, TensorFlow, or similar libraries)
- Computer vision and image processing techniques such as object detection, segmentation, and defect detection
- Machine learning / artificial intelligence model development and evaluation
- Data analysis and statistical modeling
- Imaging systems and hardware setup, including lighting, cameras, and image capture standardization
- Experimental design and model validation
- Software development and prototyping
- Operations research or process optimization

Additional helpful skills include:
- Experience with deep learning frameworks for image classification or detection
- Familiarity with dataset labeling and model training workflows
- Experience with automation, robotics, or sensor systems
- Visualization and dashboard tools (Python, Tableau, or similar) for model evaluation and operational insights

Students from disciplines such as Industrial Engineering, Computer Science, Electrical Engineering, Robotics, and Data Science would be particularly well suited for this project.

Data Access Requirement