Making use of computer visualization tools such as OpenCV, our team managed to detect defect in a manafacturing envirronment, a challenge proposed by Bobst company. We were even the only team to fully solve the problem during Lauzhack 2017 hackathon where ~250 participants where competing on various challenges.

Members

• Maxime Delisle
• Charles Gallay
• Adam Ztot
• Grégoire Clément

LauzHack is a student-run hackathon at EPFL in Lausanne, Switzerland.

Hackathons are creativity marathons, where attendees work in teams to create something exciting in a short time.

During hackatons, sponsors usually propose challenges to the attendees. We decided to go for the Bobst Challenge

Bobst is one of the world’s leading supplier of equipment and services to packaging and label manufacturers.

Develop an algorithm to rapidly identify defectives boxes in a 10.000 pictures database.

Find reliable solutions to distinguish real defects from natural process variations (position, dust, crease, etc.)

Uncover our motto based on the identified defects

Judgment criteria are algorithm reliability, process speed and the creativity of your strategy

Each team will get 1 USB 3.0 stick with the box pictures to process

Any library you can find can be used.

Our server is at your disposal to benchmark your algorithm (Linux or Windows Server).

Our engineer team will take shifts to support you night and day.

For more detailed information about Bobst and the challenge they proposed, check their tech talk slides