AI-102 Question 132
Single answerYou are designing an AI solution to analyze images provided by a retail company. They require a system that can identify if an image contains a specific type of product, such as a t-shirt, and count how many such products are in the image. Which model type should you use for this requirement?
- A
Image classification
- B
Object detection
- C
Optical character recognition (OCR)
- D
Semantic segmentation
Show answer and explanation
Correct answer: B
Explanation
The requirement to identify and count multiple instances of a specific product in an image aligns with the capabilities of object detection models. These models not only classify objects but also locate and provide bounding boxes for each detected instance, enabling accurate counting. Image classification, while useful for assigning a single label to an entire image, cannot meet the requirement of counting individual instances.
- A. Incorrect.
Image classification is used to determine the overall category or label for an entire image (e.g., 't-shirt' or 'not t-shirt'). However, it cannot count or locate individual objects within the image.
- B. Correct.
Object detection identifies and localizes multiple objects in an image, providing both the class (e.g., 't-shirt') and the bounding boxes around each detected object. This makes it suitable for identifying and counting specific products.
- C. Incorrect.
Optical character recognition (OCR) is specifically used to detect and extract text from images. It is not intended for identifying or counting objects like t-shirts in an image.
- D. Incorrect.
Semantic segmentation assigns a class label to each pixel in an image, which can help in understanding object boundaries but is not specifically designed to count discrete objects.