A company is building an AI yoga instructor that must count the number of students in a class and determine whether students perform stretches correctly by measuring limb positions and angles. Using SageMaker to extract frames from class video, which two computer-vision model types together require the least effort to implement these features? (Choose two.)

Object detection is essential for counting students. It identifies and localizes each student within the video frames, allowing for a direct count. Pose estimation is crucial for evaluating stretch correctness. It identifies key anatomical points (joints) on each student and tracks their positions and angles, enabling the AI to assess form. Image classification would only categorize the entire image (e.g., "yoga class" vs. "not yoga class") and not identify individual students or their poses. OCR is used for extracting text from images, which is irrelevant here. Image GANs are used for generating new images, not for analysis.