Computer vision methods are constantly being optimised with waves of exploration and exploitation. During exploration, exciting innovative ideas are proposed, but they have a hard time to get noticed because they do not necessarily beat the state-of-the-art (SOTA) results.  During exploitation, researchers tend to fine-tune high-performing approaches to beat the SOTA, where large efforts are needed to obtain relatively small gains. 

With the third CVF/ECCV More Exploration, Less Exploitation (MELEX) workshop we aim to publish papers in all areas of computer vision that propose explorative papers with innovative models, algorithms, and ideas that show competitive, but not necessarily the SOTA results in benchmarks.  Please visit the 1st MELEX and the 2nd MELEX workshops.

We call for unpublished innovative papers that have a history of being rejected by the main conferences because they did not achieve SOTA results. We are particularly interested in papers that have not been published, but are already receiving citations in arxiv.

Your submission must contain: 

1) manuscript (Paper format is the same as the one used for the main conference), 

2) single-page letter explaining the paper history in terms of submissions, scores and how the authors addressed the reviews. 

The submission may also contain supplementary material.

Paper submission site 

The accepted papers will be published in the ECCV proceedings.