The core objective of VehicleVision is to craft an image classification model that excels at differentiating between bicycles and motorcycles.
-
Precise Image Classification: Engineer a robust model capable of accurately categorizing images as bicycles or motorcycles.
-
Scalable Deployment: Utilize AWS Sagemaker to deploy the model in a scalable manner, accommodating varying demand.
-
Automated Workflow: Develop AWS Lambda functions to streamline data preprocessing and orchestrate their execution using AWS Step Functions.
-
Thorough Testing: Construct a comprehensive testing and evaluation framework to ensure both the model and the workflow's dependability.
-
Monitoring and Maintenance: Implement mechanisms to actively monitor model performance and identify potential anomalies.
-
AWS Sagemaker: Leveraged for model training, deployment and model monitoring, enabling scalable and efficient machine learning operations.
-
AWS Lambda: Utilized to create serverless functions for serializing images, classify images, and result filtering.
-
AWS Step Functions: Employed to seamlessly orchestrate the execution of Lambda functions, creating a coherent and automated workflow.
-
AWS S3: Utilized as a storage solution for data staging, model artifacts, and intermediate outputs during various project phases.
Prepare the dataset for model training:
- Extract data from a designated source.
- Transform data into a suitable format for training.
- Load processed data into a suitable storage system.
Train and deploy the image classification model:
- Utilize AWS's image classification algorithm for model training.
- Deploy the trained model to AWS Sagemaker endpoint.
- Configure AWS Model Monitor to track deployment performance.
Develop AWS Lambda functions and orchestrate their execution:
- Create three distinct AWS Lambda functions:
- Serialize image (
serializeImage.py
) - Classify image (
classifyImage.py
) - Result filtering (
filterInferences.py
).
- Serialize image (
- Design a workflow using AWS Step Functions to coordinate these functions (
stepFunction.json
).
- Step Function Workflow:
Thoroughly assess the workflow's effectiveness:
- Invoke the step function with test data.
- Validate successful and unsuccessful workflow executions.
- Utilize SageMaker Model Monitor insights to visualize model behavior.