Assessing integrity of shipping containers from 2D imaging - QC-150

Preferred Disciplines: Artificial intelligence: computer vision, machine learning, deep learning (Master, PhD or Post-Doc)
Company: CANSCAN Inc.
Project Length:  3-4 years (25 units)
Desired start date: Fall 2018
Location: Montreal, QC
No. of Positions: 6-8
Preferences:  None

About Company:

Canscan is a Montreal-based startup specializing in AI/Computer vision in the world of container shipping. The goal of Canscan is to find smart ways of using existing hardware available at shipping terminals to build optimization solutions for intermodal transportation. Canscan works closely with partners to create grassroot solutions to improve current container inspection processes without negativity impacting workflow or requiring expensive infrastructure upgrades.

The three pillars of Canscan:

  1. Improving health and safety
  2. Lessening environment impact
  3. Preventing commodity and asset losses

Project Description:

With a rapidly growing world population, international trade and globalization has become the foundation of the global economy. Shipping container maritime transport is the primary means by which general cargo is transported throughout the world. With over 38 million twenty-foot equivalent containers in the global fleet, the shipping container is one of the most important assets of international trade.

Built to withstand some the world’s most environmentally punishing conditions, the quality of these large metallic boxes is often overestimated resulting in the container fleet being perpetually undermaintained. The processing of huge volumes of containers at port and rail terminals require increasingly fast turnaround time leaving little to no opportunity for shipping container inspections. Container damages go unidentified leading to various harmful outcomes including health and safety issues, negative environmental impact and commodity losses.

Extreme outcomes of damaged containers remaining in circulation

  • Ship fires
  • Containers falling into the oceans
  • Hazardous goods seaping into the environment
  • Train derailments
  • Containers falling off truck chassis on highways
  • Injury or loss of life of terminal workers

The proposed project seeks to analyze two-dimensional high-definition images of shipping containers captured from video cameras located in strategic areas in terminal facilities throughout the world. The system will not only assess the types and extent of damages but also anticipate deteriorations. By moving towards a proactive/maintenance-oriented approach, not only will this project help reduce the occurrence of accidents and environmental impact, but it will also optimize logistical operations for partnering facilities, making this a win-win collaboration.

The benefits sought from this project are as follows:

  • Identify and assessing major damages with focus on those representing health and safety issues
  • Speed up turn time by limiting worker inspections to exception cases
  • Anticipate deteriorations, encouraging pro-active maintenance practices and reducing the need for international transportation of empty damaged containers to repair facilities

Research Objectives/Sub-Objectives:

  • Pre-Processing of videos, segmentation of images
  • Cleaning / denoising / compensating
  • Cloud-computing architecture
  • Recognition of container labelling:

a. Alpha-numeric digits
b. Recognizing hazardous placards (signs)
c. Traffic Sign Recognition

  • Discern damaged from not damaged
  • R&D to recognize container damage types such as rust, perforations, bulging, cracks, cuts, poor quality repairs, etc.
  • Integration through API into collaborative partner operating system
  • Testing and modifications

MAJOR TECHNICAL CHALLENGES

  • Varying environment parameters: lighting, proximity to camera, obstructions (snow/ice/dirt/fog)
  • Identifying depth related damages from 2D imaging

Methodology:

  • HD imaging from clients/partners in manufacturing (rail and port terminals)
  • Cloud computing and data storage
  • Usage of existing open source computer vision libraries
  • Deep learning techniques
    • Usage of existing libraries: OpenCV and Python programming
      • Fast R/CNN / YOLO
      • Usage of existing libraries: OpenCV, SIFT, Sobel filters, Hough lines
    • DELF: DEep Local Features
    • Crowdsourced image labelling

Expertise and Skills Needed:

General

  • Leadership skills: willing to take charge of a small team to achieve project milestones
  • Willing to learn about intermodal transportation and take part in industry events
  • Passionate about green technology and industry 4.0
  • Knowledge of both French and English will be a plus

Tech

  • Computer vision
  • Machine learning, with focus on image analysis recognition
  • Programming skills (Python, Java, etc.)
  • Extensive experience with NN training frameworks like PyTorch, Tensorflow, etc.
  • Usage of existing libraries: Open CV, Visual SFM, Point cloud library (PCL), Geometry factory, Cloud compare, Unity,Etc.
  • Experience/Previous Studies with DEep Local Features (DELF)
  • Knowledge in cloud computing (AWS) will be a plus
  • Experience with crowd sourced image labeling a plus

For more info or to apply to this applied research position, please

  1. Check your eligibility and find more information about open projects.
  2. Interested students need to get approval from their supervisor and send their CV along with a link to their supervisor’s university webpage by applying through the webform or directly to Gabriel Garcia-Curiel at ggarciacuriel(a)mitacs.ca.
Program: