3D Model Identification - BC-360
Preferred Disciplines: PhD students experienced in image processing and machine learning
Project length: 4 to 6 months
Desired start date: ASAP
Location: Downtown Vancouver. BC
No. of Positions: 2
Preferences: Language: English. UBC, SFU
Company: Kabuni Technologies Inc
When you think of 3D printing? There is no category king.
Combining artificial intelligence, artificial reality and 3D Printing to Build the World’s Leading Retail Experience
Print the Future is establishing brand credibility and recognition as the premier name associated with the 3D print industry. We are uniquely positioned to use AI to design, AR for visualization and 3D printing to bring to life, in order to be the category's dominant force, synonymous with the industry itself, and become a new category king.
The company is developing an intelligent engine (named Gurru) that will unlock historical design data to generate new designs based upon specific user needs.
As an example, if a customer wanted to find a chair, and used Gurru to help them find it, Gurru would look at all previous data on chairs made and sold. Gurru would also take into consideration that customer's unique attributes (these may be personal characteristics like weight, health, malady, age etc) in order to present an entirely new chair design based on their specific requirements. Gurru’s recommendation engine would also take into account other factors (like geographical location and possibly ethnic backgrounds) to predict styles and taste and present them with a product tailored to them. This design would ultimately be unique to the customer based on information outside of current standard features like colour and size.
This selected item will then be 3D printed by the Company at one of its locations across North America.
The main development will be for lifestyles designs, which would be mostly furniture and household items. The selected product will be designed by the AI component of Gurru.
Background and required skills
In this project, we are looking to extract a set of features from our 3D models that can be used as fingerprints. We would like to use the extracted features to design and train classification algorithms that measure the similarity between the 3D model and find similar 3D models.
As a part of this project, we would like to extend our dataset of 3D models and also add descriptions to the items that are not labeled via supervised and unsupervised methods.
- Support Vector Machines
- Convolutional Networks
- Recurrent Neural Networks
- Deep Boltzmann Machines
- Convolutional Deep Belief Networks
Expertise and Skills Needed:
- Previous experience in image processing and machine learning (feature extraction, image classification, adversarial training, image captioning)
- Understands the balance between delivering a MVP to meet customer needs today while anticipating platform scalability and growth
- Self-starter, organized, and details-oriented with strong analytical skills
- Familiar with NLP (building natural language understanding, image captioning)
- Familiarity with the open-source scientific computing frameworks like Torch, TensorFlow, Pytorch and Theano
- Professional experience with Python
- Familiarity with AWS cloud computing
- Experience in big data framework such as Spark and Hadoop
- Excellent presentation and communication skills
For more info or to apply to this applied research position, please