Samples
Street View Insights can help you analyze imagery datasets using Vertex AI Colab Enterprise. The following examples showcase various capabilities.
Environment setup
This workshop is designed to be run in Vertex AI Colab Enterprise. Follow the instructions below to import the tutorial notebooks into your environment:
- Import Notebook: In Colab Enterprise, select File > Import notebook and choose the "By URI" option.
- Copy & Paste: Copy the Import URI provided within each module card below and paste it into the import dialog.
- Rename File (Recommended): To avoid conflicts, consider renaming the imported notebook file, for example by appending your username to the start of the filename (e.g.,
{USERNAME}_filename.ipynb).
Basic modules
These modules cover the fundamental workflows for getting started with Street View Insights.
1. Understand your dataset
Introductory notebook to explore and visualize the imagery dataset structure and associated metadata. Import URI:
https://github.com/googlemaps-samples/insights-samples/blob/main/street_view_insights/notebooks/Understand_your_dataset/Understand_your_dataset.ipynb
2. Utility pole analysis
Core analysis workflows for identifying and categorizing utility poles based on their visual features. Import URI:
https://github.com/googlemaps-samples/insights-samples/blob/main/street_view_insights/notebooks/Utility_pole_analysis/utility_pole_basic_analysis.ipynb
3. Classify road signs
Classify road signs found in imagery, such as Stop, Yield, and Speed Limit signs. Import URI:
https://github.com/googlemaps-samples/insights-samples/blob/main/street_view_insights/notebooks/classify_road_signs/classify_road_signs.ipynb
Advanced modules
These modules cover more complex analyses and techniques, including AI-powered features like few-shot learning and code execution.
4. Object detection with few-shot learning
Detect objects in images by training a model on just a few examples—ideal for identifying rare or custom objects.
See: Few-Shot Examples
Import URI:https://github.com/googlemaps-samples/insights-samples/blob/main/street_view_insights/notebooks/Object_detection_with_few_shot_learning/Object_detection_with_few_shot_learning.ipynb
5. Attachment detection
Bounding box (bbox) detection for various pole attachments, such as transformers, crossarms, and insulators. Import URI:
https://github.com/googlemaps-samples/insights-samples/blob/main/street_view_insights/notebooks/bbox_detection_of_attachments/bbox_detection_of_attachments.ipynb
6. Lean angle detection
Advanced analysis to calculate the lean angle of poles from imagery, which can be used to assess pole stability.
See: Code Execution
Import URI:https://github.com/googlemaps-samples/insights-samples/blob/main/street_view_insights/notebooks/Lean_angle_detection_of_pole/Utility_pole_lean_angle_detection.ipynb
7. Utility pole height measurement
Measure the height of utility poles from imagery using object detection and geometric analysis.
See: Structure Prompts
Import URI:https://github.com/googlemaps-samples/insights-samples/blob/main/street_view_insights/notebooks/Utility_pole_measure_height/Utility_pole_measure_height.ipynb
8. Evaluation metrics
Evaluate model performance and analysis results using industry-standard computer vision metrics. Import URI:
https://github.com/googlemaps-samples/insights-samples/blob/main/street_view_insights/notebooks/eval/eval.ipynb
9. Image quality analysis
Assess image quality based on factors like blur and lighting to ensure suitability for computer vision tasks.
See: Code Execution
Import URI:https://github.com/googlemaps-samples/insights-samples/blob/main/street_view_insights/notebooks/Image_quality_analysis/Image_quality_analysis.ipynb