Over the past decade, there has been tremendous progress in the field of computer vision. Thousands of research papers are published yearly, with many obtaining state of the art results on established benchmarks.

With this frenetic activity, it can be challenging to get a sense of how well popular models perform on data you may be working with. The PRIOR team has found that exploring a model’s qualitative behaviour can provide insights that are hard to get purely from tracking quantitative metrics. Online demos offer researchers a quick and easy way to perform a qualitative error analysis on small samples of their data and evaluate whether a given model may be useful for downstream tasks.

This page showcases a number of accessible models achieving state of the art (or near state of the art) results on many popular computer vision tasks. We hope you find this useful!

This page is a work in progress. We will continue to add state of the art models as they get released. Stay tuned!

This demo page is built and maintained by PRIOR and colleagues at the  Allen Institute for AI. Our team seeks to advance computer vision to create AI systems that see, explore, learn, and reason about the world.