Computer Vision Engineer

Role: Data

Salary: $175,000 - $175,000

Job Reference: 190467

*This position is unable to provide sponsorship, Candidate must be US a citizen or green card holder*

 

For this role the ideal candidate will:

-Be proficient 3D image-based reconstruction experience, C++, Python

 

The computer vision engineer will be an expert on building large-scale image retrieval systems that handle diverse environments and changes in scene appearance over time. Design and implement production-ready solutions that can scale to billions of images.

 

Tasks

Sample tasks that would be completed during the first year:

  • Image retrieval
  • Determine what techniques we should be using to build up a searchable database of imagery
  • Find sets of similar images that could be used for 3D reconstruction
  • Given a patch of pixels, determine if that patch had been seen elsewhere (identification of railcars, objects, etc)
  • Appearance change over time
  • Enhance our feature matching, image retrieval, and 3D reconstruction strategies to deal with changing scene appearance over time (how to know if it's the same scene even when the appearance has changed)
  • Wide baseline matching
  • Implement techniques to improve our ability to match visual features when there is a large difference in viewing direction
  • Feature tracking and motion detection
  • Work on techniques to identify patterns of motion in images (haul trucks driving around a site, pallets being loaded, etc)
  • Assist in developing new 3D reconstruction strategies and applications
  • Installed cameras for pallets (volume estimation, counting, motion tracking, etc)
  • Large-scale SfM
  • Possible Stockpile Reports tasks:
  • Optimize performance of our C++ and Python pipelines

Sample Milestones

  • After 1 month
  • Build prototype to perform image retrieval using 100K images from EveryPoint datasets
  • Teach us the best practices when performing large-scale image retrieval
  • After 2 months
  • Refine the image retrieval prototype to identify the same, isolated object within a large dataset of images
  • Update our processing pipelines to enable easier future use of image retrieval (e.g. use a consistent feature vocabulary, save the visual words for an image, etc)
  • Teach us how to maintain high retrieval performance under changing scene conditions and appearance
  • After 3 months
  • Create prototype for motion detection logic to identify the moving parts of a scene, ignore outliers, corroborate with other views, etc
  • Create prototype for wide baseline matching

For more information about this role please apply, or email Micah Rapisand on mrapisand@pg-rec.com who will be able to provide more details on the opportunity.

Many thanks, Premier Group

Share article:

Consultant

Micah Rapisand

mrapisand@pg-rec.com

Similar jobs:

Woking

Product Manager, Growth

Location:

New York

Salary:

$160,000 - $160,000

Read more

Woking

Product Manager, Growth

Location:

New York

Salary:

$160,000 - $160,000

Read more

Woking

Product Manager, Growth

Location:

New York

Salary:

$160,000 - $160,000

Read more

Share article: