A beauty-supply retailer wants an hourly report of store visitors derived from years of security video. The report must categorize visitors by hair style and hair color. Which approach requires the least development effort?

The most efficient approach for this task is to use an object-detection model to locate hair and then a ResNet-50 classifier. Object detection models are designed to identify and localize specific objects (like hair) within an image, providing bounding boxes that can be cropped. ResNet-50 is a pre-trained convolutional neural network (CNN) widely used for image classification, making it suitable for categorizing hair style and color with minimal additional training data or architectural changes. Using XGBoost for image classification is less suitable because it is a tree-based model that typically performs better on tabular data and would require extensive feature engineering from image data, increasing development effort. Semantic segmentation, while more precise in outlining hair regions, is overkill for this problem and requires more complex model architectures and annotation efforts than simple object detection, thus increasing development time.