Great thanks to the team @ruojin.bsky.social, Hanyu Chen, Jeffrey Byrne and @snavely.bsky.social

🌟Doppelgangers++ outperforms existing SfM pipelines and previous works with improved precision and recall, even in out-of-domain tests, whilst requiring less hyperparameter tuning.

🔍Key highlights:
- Transformer-based classifier leveraging 3D-aware features from MASt3R;
- Features VisymScenes, a new dataset of daily, challenging scenarios collected worldwide;
- Seamless integration with SfM pipelines like COLMAP and MASt3R-SfM.

Introducing Doppelgangers++! 🚀 An enhanced pairwise image classifier that tackles visual aliasing (doppelgangers) to improve 3D reconstruction accuracy across diverse, real-world scenes. 🌍✨
🔗Project page: bit.ly/3VAPMJc. Code is also available.