GCP Cloud Digital Leader Practice Question

AutoML Vision is specifically designed for customers who have their own labeled images but limited machine-learning expertise. It automates the processes of model architecture selection, training, and hyper-parameter tuning, letting the company build a high-quality custom classifier in hours or days without writing training code. The pre-trained Cloud Vision API is fast to adopt but cannot detect product-specific defects that were not part of Google's public training data. BigQuery ML focuses on structured, tabular data and does not train models directly on raw images, so it is unsuitable here. Building a TensorFlow model from scratch on Vertex AI Workbench can achieve similar accuracy but requires significant ML skills and effort that the company does not have, delaying time to value.