AI-102 Question 130
Single answerYou are building a solution for a retail client that requires analyzing security camera footage to detect shoplifting incidents. The client wants to identify specific objects such as 'bags' and 'items being removed from shelves' in real-time. Which type of AI model should you choose to meet the client's requirement?
- A
Use an image classification model to categorize the entire image as shoplifting or not shoplifting.
- B
Use an object detection model to identify and locate specific objects like 'bags' and 'items being removed from shelves' in the footage.
- C
Use an image classification model to detect the presence of shoplifting-related objects in the image.
- D
Use a text analysis model to analyze the scene descriptions for shoplifting incidents.
Show answer and explanation
Correct answer: B
Explanation
The client requires detecting and locating specific objects ('bags' and 'items being removed from shelves') in real-time video footage. Object detection models are specifically designed for this purpose as they can identify and localize multiple objects within an image. Image classification models, on the other hand, provide a single label for the entire image and cannot localize objects, making them unsuitable for this requirement.
- A. Incorrect.
Image classification models are designed to assign a single label to an entire image, such as 'shoplifting' or 'not shoplifting'. However, this approach cannot localize or identify individual objects within the image, making it unsuitable for the given requirement.
- B. Correct.
Object detection models are capable of detecting specific objects in an image and providing their locations (bounding boxes). This makes them ideal for identifying objects like 'bags' and 'items being removed from shelves,' as required by the client.
- C. Incorrect.
Image classification models can detect the presence of certain objects but cannot provide their locations or distinguish between multiple objects in the image. This makes them inadequate for scenarios that require object localization.
- D. Incorrect.
Text analysis models are used for processing text data, such as extracting insights from written descriptions. They are unrelated to the visual analysis of images or videos, making them irrelevant to the given scenario.