Learnable Triangulation for Deep Learning-based 3D Reconstruction of Objects of Arbitrary Topology from Single RGB Images.
Tarek Ben CharradaHedi TabiaAladine ChetouaniHamid LagaPublished in: CoRR (2021)
Keyphrases
- deep learning
- arbitrary topology
- three dimensional
- multiple objects
- image data
- image features
- input image
- image registration
- complex scenes
- object recognition
- bounding box
- image retrieval
- computer vision
- test images
- partial occlusion
- target object
- d objects
- keypoints
- geometric constraints
- multiple images
- graph cuts
- image processing