11-30, 11:35–12:05 (Europe/Amsterdam), Auditorium
Tasks related to computer vision don’t have an easy learning curve - there are many nuances, it’s in some cases subjective and the data is not always there and ready to be used.
In this talk, you will see an example of how to use your own domain-specific data to outperform pre-built market solutions. We will explore some approaches for generating ground truth data for a background removal task, and how to use it for fine-tuning open source models with limited computational power.
Background removal is a process in image editing and computer vision where the main subject of an image is isolated from its background. Among many spheres where it is applied, product design is one where it is heavily relied upon.
The talk is catered primarily towards data scientists and Python enthusiasts who would love to know more about practical computer vision problems. During this presentation, you will learn about our approach to how background removal was tackled in a print-on-demand company with no ground truth data, see how initial research can be done (even without a pricey GPU-based instance), get to see iterative improvements in the results and understand how eventually domain-centric training results can compare to external providers.
No previous knowledge expected
Data Scientist at Printify