PROFESSIONAL-MACHINE-LEARNING-ENGINEER · Question #250
You work for a pet food company that manages an online forum. Customers upload photos of their pets on the forum to share with others. About 20 photos are uploaded daily. You want to automatically…
The correct answer is A. Send user-submitted images to the Cloud Vision API. Use object localization to identify all objects. For automatically and in near real-time detecting animals in a small volume of uploaded photos with minimal development time and cost, leverage the pre-trained Google Cloud Vision API.
Question
Options
- ASend user-submitted images to the Cloud Vision API. Use object localization to identify all objects
- BDownload an object detection model from TensorFlow Hub. Deploy the model to a Vertex AI
- CManually label previously submitted images with bounding boxes around any animals. Build an
- DManually label previously submitted images as having animals or not. Create an image dataset
How the community answered
(25 responses)- A80% (20)
- B4% (1)
- C4% (1)
- D12% (3)
Why each option
For automatically and in near real-time detecting animals in a small volume of uploaded photos with minimal development time and cost, leverage the pre-trained Google Cloud Vision API.
The Cloud Vision API provides pre-trained models capable of object detection and classification, including identifying animals. It's a fully managed service, requiring minimal development time and deployment effort, and is cost-effective for a low volume of 20 photos daily, perfectly matching the requirements for speed, cost, and minimal development.
Downloading a model from TensorFlow Hub and deploying it to Vertex AI requires more development effort, model management, and potentially higher costs compared to using a pre-trained, managed API for a simple task and low volume.
Manually labeling images and building a custom object detection model involves significant time and cost for data annotation, model training, and deployment, which contradicts the goal of minimizing time and cost.
Manually labeling images for image classification also incurs significant data labeling and training costs/time, even if it's simpler than object detection, making it less optimal than using a ready-made API for this specific low-volume task.
Concept tested: Pre-trained APIs vs. custom ML models, cost-effectiveness
Source: https://cloud.google.com/vision/docs/object-localization-tutorial
Topics
Community Discussion
No community discussion yet for this question.