Top-tier ML and vision research, 2016 to 2026
Select a stream or a bar to go one level deeper.
108,967 accepted papers, 2016 to 2026
Select a stream or a bar to go one level deeper.
Universities, labs and companies ranked by accepted papers. Pick a venue above, a year range and, if you like, a field.
Top 10 above, ranked among the 20 listed, each year.
Papers: a paper counts once for every institution with at least one author on it, so a paper by Tsinghua and Google adds 1 to each. Fractional: each author is 1/(number of authors) of the paper, split evenly when an author lists two institutions, so a paper adds up to 1 in total. With 3 Tsinghua authors and 1 Google author, Tsinghua gets 0.75 and Google 0.25. This favours institutions that lead papers over those on many large collaborations.
Organizations counts labs under their parent (Google DeepMind and Google Research under Google). Institutions come from OpenAlex where the paper is matched there, otherwise from the affiliations printed on the paper, cut down to the organization. Papers without known affiliations are left out; the line above the ranking says how many that is.
An AI paper atlas: a research map of machine learning, vision and AI papers. Describe what you plan to try: scime places it among every accepted paper from ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) and shows its area, its approach and the closest existing work.