Thin-object detection of utility hardware: wires, guys, and anchors - BC-1037

Project type: Research
Desired discipline(s): Engineering - computer / electrical, Engineering, Computer science, Mathematical Sciences, Mathematics
Company: Anonymous
Project Length: 4 to 6 months
Preferred start date: 11/02/2026
Language requirement: English
Location(s): Richmond / Vancouver, BC, Canada
No. of positions: 1
Desired education level: Master'sPhD
Open to applicants registered at an institution outside of Canada: No

About the company: 

The partner organization (ANONYMOUS) is an established Canadian electrical utility consulting firm based in British Columbia, with a team of approximately 95 professionals. The company provides engineering services to major Canadian utilities across transmission, distribution, substation, and telecommunications infrastructure. A significant part of its practice is distribution pole replacement design: assessing end-of-life wood poles and producing the engineering design required to replace them safely and to standard. The company is investing in an applied AI/R&D program that uses computer vision and machine learning to automate the interpretation of field photographs of utility poles, so that engineering design information currently extracted manually can be pre-filled automatically with calibrated confidence scores and clear flags for designer review. Interns joining this program work with a real production dataset of field imagery and design records, alongside the company's internal ML, platform, and engineering teams, with a direct path from research results to deployment in a production design workflow.

Describe the project.: 

What is the project about / main goal: This project treats overhead wires, guy wires, and anchors as one coherent thin-object detection problem — the annotation protocols, modeling approaches, counting logic, and failure modes overlap enough that a combined treatment is stronger than separate efforts. The goal is to count and type this hardware from pole photographs with calibrated confidence and a designer-review fallback for occluded or ambiguous cases. Thin objects are frequently occluded by vegetation, blend into cluttered backgrounds, and are easy to confuse across viewing angles, making this a challenging and technically coherent research problem.
Main tasks to be performed by the candidate:
• Develop a shared polyline / keypoint / instance-mask annotation protocol for thin objects
• Count individual overhead wires, triplex, and duplex cables, classifying each as secondary or service where visible
• Count guy wires and anchors per pole, linking top attachment points to ground anchors where possible
• Evaluate edge-aware segmentation, keypoint tracing, and query-based instance detection methods
• Handle occlusion, perspective distortion, and visually similar cable types; stretch: guy/anchor type classification and sidewalk guy assemblies
Methodology/techniques to be used: Thin-structure segmentation and keypoint/instance detection; targets: wire counting MAE ≤ 0.5, guy/anchor count exact-match ≥ 0.60, type macro-F1/accuracy ≥ 0.65 where labels are reliable.

Required expertise/skills: 

Thin-structure segmentation; keypoint / instance detection; strong PyTorch.
Optional: assets
Production dataset of utility-pole field imagery with linked engineering design records; internal labeling platform and labeling pipeline; existing trained baseline models to build on; cloud GPU compute; day-to-day co-supervision by the company's internal ML and platform team.