DDD20 DAVIS event camera driving dataset
by Tobi
It's finally available! We are proud that our DDD20 paper was accepted to the Intelligent Transportation Systems Conference. Check out the DDD20 website and arxiv prepublication for lots of cool information about this unique end-to-end event camera dataset.
Now we added the full ITSC talk on DDD20.
This paper shows that fusing events with frames produces higher steering prediction than either modality alone.
By sharp contrast to previous results, the DVS events supplement the APS frames, which are by themselves somewhat better than DVS frames for predicting steering.
Congratulation to PhD student Yuhuang Hu for this useful contribution to the research community.