Image Classification Model for Product Defects

About Role

Develop a proof-of-concept computer-vision model that classifies product images into approved and defect categories. The project includes data review, labeling guidance, model training, evaluation, and a simple demonstration interface or API. We need honest performance reporting and a clear plan for improving the model with more data. The successful freelancer will be expected to work independently, share progress at agreed checkpoints, and raise questions early when a requirement needs clarification. All submitted work should be original, well organized, and ready for practical use after reasonable review.

Requirements

• Review the image dataset and identify labeling or imbalance issues.
• Suggest a practical annotation and data-splitting strategy.
• Train and compare suitable classification approaches.
• Report precision, recall, confusion matrix, and important failure cases.
• Create a simple inference script or API demonstration.
• Document how new images can be added for future training.