Answer: Import the labeled images as a managed dataset in Vertex AI and use AutoML to train the model.

The question asks for developing an image classification model using a large dataset of labeled images in Cloud Storage. Option C is optimal because Vertex AI AutoML is specifically designed for such scenarios, automatically handling model architecture selection, hyperparameter tuning, and training with minimal setup. It directly supports importing labeled images from Cloud Storage as managed datasets and is ideal when the primary goal is model training without complex pipeline requirements. While options A and B (using Kubeflow or TFX pipelines) are technically feasible, they introduce unnecessary complexity for a straightforward image classification task. Option D is unsuitable as converting images to tabular format for BigQuery ML is inefficient and loses spatial information critical for image classification. The community discussion shows strong support for C (60% of answers), with key points being AutoML's simplicity and direct applicability, though some debate exists about dataset size limitations (AutoML supports up to 1 million images, sufficient for most 'large' datasets).

Author: LeetQuiz Editorial Team