SHARP2023 Challenge

Register Early for dataset access!

Join the 4th edition of the SHARP (Solving CAD History and pArameters Recovery from Point clouds and 3D scans) Challenge! The SHARP Challenge is organized in conjunction with the ICCV 2023 and the SHARP Workshop. The 3-Track SHARP Challenge promotes a competition to encourage the development of new methods for recovering the history and parameters of Computer-Aided Design (CAD) models from 3D scanned data. Each Challenge track employs a different version of the CC3D dataset. The CC3D dataset is derived from open CAD repositories such as 3D Content Central and contains over 50k pairs of CAD models and their corresponding 3D scans.

A total of 8000€ will be awarded in cash prizes by Artec3D to the winners!

Further documentation and I/O Python routines

Available on our SHARP 2023 Gitlab Repository

The tracks available on Codalab!

Track1 – Track2 – Track3

Timeline

Tracks

Track 1: Parametric sharp edge inference
- Task: Simultaneously detect edges (as parametric curves or straight line-set) & binarily classify detected edges as either sharp or not.
- Dataset: CC3D-PSE (includes Boundary-Representation (B-Rep) edges, degree of sharpness annotations & and binary edge labels)
Track 2: Segmenting scans into Boundary-Representation (B-Rep) face memberships
- Task: Segment & assign vertices of 3D scan into CAD surface primitives (equivalent to B-Rep faces)
- Dataset: CC3D-Brep (includes 3D scans and B-Rep face annotations)
Track 3: Segmenting scans into CAD operations types and ordered steps
- Task: Segment scan vertices into ordered steps of CAD modeling/design history & operation types associated with them.
- Dataset: CC3D-ops (includes annotations on CAD operation types i.e., extrusion, revolution, chamfer, etc.)

Registration

In order to participate in one or more tracks, including for the online option, participants need to express their interest by simply using the following contact form.

Participation to multiple tracks is possible by selecting multiple choices for the challenge in the form.

A data license agreement to access the datasets will be automatically sent to all participants. The data license agreements need to be signed and returned by email.

Access to download the training data after signing license agreements by all parties.

For any enquiries, please contact us on Shapify3D (at) uni.lu