Blue Flamingo

This paper presents a novel Mutual Learning to Adapt model(MuLA) for joint human parsing and pose estimation. NuLA predicts dynamic task-specific model parameters by incorporating information from their counterparts.

This paper proposes a novel Pose Partition Network (PPN) that learns joint detector and dense regressor simultaneously to address the challenging multi-person pose estimation.

This paper proposes a novel Parsing Induced Learner to exploit parsing information to effectively assist pose estimation by learning to fast adpat the base pose estimation model.

This paper designs a Contextual Refinement Unit(CRU) to implement a new Hierarchical Contextual Refinment Network(HCRN) to predict positions of joints from easy to difficult.

This paper presents an approach for generating face animations from large unorganized image collections of the same person, by creating a graph with faces as nodes, and similarity as edges, and solving for walks and shortest paths on this graph.

This paper provides an image deformation method based on Moving Least Squares using various classes of linear functions including affline, similarity and rigid transformation.

This paper parameterizes a map over a halfway domain with a few user-drawn points to align corresponding image elements automaicly, and improve the morphs by optimizing quadratic motion paths and extending content beyond the image boundaries.

This paper propose to apply 3D CNNs to learn the patio-temporal correlation features jointly from low-level to high-level layers for trafic data and design an end-to-end structure named MST3D.

This paper propose the 'double attention block' to aggregate and propagate informative global features from the entire spatio-temporal space of input images/videos, enabling subsequent convolutional layers to access features from the entire space efficiently.