A logistics company installed in-vehicle cameras to monitor drivers and wants to reduce distraction-related accidents with minimal operational effort. Which solution requires the least operational work to identify driver distractions?

The most operationally efficient solution is using Amazon Rekognition's eye gaze direction detection. Rekognition is a fully managed service that offers pre-trained computer vision capabilities, including facial analysis. This means you can integrate it directly with your camera feeds to detect eye gaze without needing to build, train, or manage custom machine learning models. This significantly reduces operational overhead. Customizing an AI model with Amazon SageMaker would require extensive data labeling, model training, and ongoing management, which is high operational effort. Integrating a third-party system, while potentially effective, introduces additional complexity in terms of integration, maintenance, and vendor management, increasing operational work compared to a native AWS service. Amazon Comprehend is designed for natural language processing (NLP) and text analysis, not for analyzing video feeds from in-vehicle cameras, making it unsuitable for this task.