Which method quickly builds an initial vehicle damage detection model on Google Cloud?

You work for an auto insurance company. You are preparing a proof-of-concept ML application that uses images of damaged vehicles to infer damaged parts. Your team has assembled a set of annotated images from damage claim documents in the company’s database. The annotations associated with each image consist of a bounding box for each identified damaged part and the part name. You have been given a sufficient budget to train models on Google Cloud. You need to quickly create an initial model. What should you do?

Download a pre-trained object detection model from TensorFlow Hub. Fine-tune the model in Vertex AI Workbench by using the annotated image data.
Train an object detection model in AutoML by using the annotated image data. Source Reference Answer
Create a pipeline in Vertex AI Pipelines and configure the AutoMLTrainingJobRunOp component to train a custom object detection model by using the annotated image data.
Train an object detection model in Vertex AI custom training by using the annotated image data.

Community Votes

100%

100% of anonymous learners picked answer B. Votes are pick records left by other test-takers — they are not the verified answer.

Community Insight

The question tests matching project constraints to a training approach, and the trap is over-engineering with custom training, pipelines, or manual fine-tuning when AutoML is the intended fast path.

For a proof-of-concept object detection model using annotated bounding-box images, Vertex AI AutoML is the fastest path to a quality initial model with minimal code. The community unanimously (100%) agrees that 'quickly create an initial model' is a direct keyword signal for AutoML.

Choosing A (fine-tuning a pre-trained TensorFlow Hub model) because transfer learning sounds fast, but it requires custom code, architecture decisions, and more ML expertise than AutoML's managed, automated training.

Community Discussion (7 comments)

louisaok 👍 1 Selected: B

>>" You have been given a sufficient budget to train models on Google Cloud" it is rare to see a company give enough money to run a mission-critical project.

Foxy2021 👍 1

My vote is B

VinaoSilva 👍 2 Selected: B

quickly create an initial model = automl

pinimichele01 👍 2 Selected: B

went with B

edoo 👍 2 Selected: B

By doing B we are doing D. I suppose B in more specific about the model and thus "more" correct? Thoughts?

ddogg 👍 1 Selected: B

B makes the most sense, data is already labelled and a pretrained model may not fit for this specific case

pikachu007 👍 4 Selected: B

Speed: AutoML excels in creating high-quality models with minimal code and setup, significantly accelerating model development. Ease of use: It provides a user-friendly interface and automates many aspects of model training, making it accessible even for those without extensive ML expertise. Automatic optimization: AutoML automatically handles hyperparameter tuning, feature engineering, and architecture selection, reducing manual effort and expertise required. Custom object detection: It supports custom object detection tasks, directly addressing the need to identify damaged parts in images.

Comments & Corrections

No comments yet — spotted an error or have a note? Share it below.

Expert Analysis

Why the Answer Is Correct

AutoML on Vertex AI is purpose-built for producing high-quality custom models quickly with minimal code, which directly matches the requirement to "quickly create an initial model" for a proof of concept. The dataset is already annotated with bounding boxes and part labels, which is exactly the input format AutoML image object detection expects. Comment [1] highlights AutoML's speed, ease of use, and automatic hyperparameter tuning, while comment [2] distills the keyword mapping: "quickly create an initial model = AutoML." Since the budget is stated as sufficient, AutoML's managed training cost is not a blocker, making B the clear best answer.

Why the Other Options Are Wrong

Option A requires downloading and fine-tuning a pre-trained TensorFlow Hub model inside a Workbench notebook, which demands custom training code and more iteration time; comment [5] notes that a generic pre-trained model may not fit this specific vehicle-damage domain. Option C wraps AutoML training inside Vertex AI Pipelines, adding orchestration overhead that is valuable for production MLOps but unnecessary for a one-off proof of concept. Option D (Vertex AI custom training) offers maximum control but forces you to choose an architecture, write training code, and tune hyperparameters yourself, slowing initial delivery. Comment [3] captures the nuance that AutoML training is technically a form of training, but B is "more specific" and therefore the better answer.

Community Comment Notes

The community is fully aligned: 100% of votes went to B, with no dissenting answers in the comments. Comment [1] provides the strongest rationale, citing AutoML's minimal setup, automated feature engineering, and hyperparameter optimization. Comment [2] reinforces the exam keyword pattern that phrases like "quickly" and "initial model" signal AutoML. Comment [4] points to the budget clue as a hint that managed training is affordable, and comment [5] adds that pre-trained generic models may not transfer well to domain-specific damage detection.

Official Reference

Exam Strategy

Map urgency keywords to managed services: 'quickly,' 'initial model,' 'minimal effort,' or 'proof of concept' almost always points to AutoML, while 'custom architecture' or 'full control' points to custom training. Remember that Vertex AI Pipelines add orchestration value for production workflows, not for one-off PoC training runs. Always pick the simplest option that satisfies every stated constraint in the scenario.

Related Analysis

Practice All PMLE Questions

Access 65 questions with complete answers and detailed explanations.

View Full PMLE Practice Test →

← Back to PMLE Study Guide