endorsedCompleted

CV ML Models and MLOps for Electric Utility Field Asset Inspection

Services Covered:

AI and Machine Learning • Cloud Consulting and Services

Industry

Utilities

Duration

12 months

Budget

200K - 999K

Start Date

1 June 2020

Team Size

6-10 Employees

Engagement Model

Onshore

Project Capability Score (PCS)

69/100

Good

Project Capability Score (PCS) estimates how capable ThirdEye Data is of successfully delivering a project like this, based on its past experience, track record, and ability to handle similar work. Learn more about PCS

Client

Southern California Edison

Rosemead, United States of America

Project Summary

ThirdEye Data developed computer vision machine-learning models for an electric utility's field asset inspection program and provided MLOps architecture consulting. They productionized models into an ML pipeline on Microsoft Azure and supported the client through changing requirements, delivering most models on schedule while advising on roadmap and continuous improvement.

Key Challenges

Develop ML models to automate image quality assessment for field asset inspections
Consult on and design MLOps architecture and a roadmap for continuous improvement
Handle changing data formats and technical complexity of novel models

Project Deliverables

Computer vision ML models for image quality assessment for field asset inspection
Models to identify assets and detect asset components
Models to assess angles, coverage, occlusion, and blur
Productionized ML pipeline on Azure

Project Solution

ThirdEye Data developed multiple computer vision ML models to automate image quality assessment for field asset inspection, including asset identification, component detection, and occlusion/blur detection. They productionized the working models into an ML pipeline on Azure and provided MLOps architecture consulting and a roadmap for continuous improvement.

Project Outcome

Delivered 4 of 6 models on time meeting scope and quality targets
Two models did not meet expected performance and were not deployed
Team adapted to changing requirements (e.g., pole tag format changes) and met performance metric targets for delivered models

Platforms

Cloud

Tech Stack

Azure

Client Endorsement

Overall Review Rating

4.75

Timeliness

Cost Rating

Willing to Refer

Quality of Deliverables

“ThirdEye Data was adaptive to the client's changing requirements and successfully met model performance metric targets. The team delivered most of the ML models on time, and they were very collaborative and supportive throughout the engagement. Moreover, they were committed, honest, and open-minded.”
Kumar Mankala
Former Advisor of AI/ML Program & Products
Date: 06/07/2023