Abstract-In this study, we show that landmark detection or face alignment task is not a single and independent problem. Instead, its robustness can be greatly improved with auxiliary information. Specifically, we jointly optimize landmark detection together with the recognition of heterogeneous but subtly correlated facial attributes, such as gender, expression, and appearance attributes. This is non-trivial since different attribute inference tasks have different learning difficulties and convergence rates. To address this problem, we formulate a novel tasks-constrained deep model, which not only learns the inter-task correlation but also employs dynamic task coefficients to facilitate the optimization convergence when learning multiple complex tasks. Extensive evaluations show that the proposed task-constrained learning (i) outperforms existing face alignment methods, especially in dealing with faces with severe occlusion and pose variation, and (ii) reduces model complexity drastically compared to the state-of-the-art methods based on cascaded deep model.
In the U.S., 30% of adults suffer joint pain, most commonly in the knee, which severely limits mobility and is often attributed to injury of cartilage and underlying bone in the joint. Current treatment methods such as microfracture result in less resilient fibrocartilage with eventual failure; autografting can cause donor site morbidity and poor integration. To overcome drawbacks in treatment, tissue engineers can design cell-instructive biomimetic scaffolds using biocompatible materials as alternate therapies for osteochondral defects. Nanofibrous poly (L-lactic acid) (PLLA) scaffolds of uniform, spherical, interconnected and well-defined pore sizes that are fabricated using a thermally-induced phase separation and sugar porogen template method create an extracellular matrix-like environment which facilitates cell adhesion and proliferation. Herein we report that chondrogenesis and endochondral ossification of rabbit and human bone marrow stromal cells (BMSCs) can be controlled by scaffold pore architecture, particularly pore size. Small-pore scaffolds support enhanced chondrogenic differentiation in vitro and cartilage formation in vivo compared to large-pore scaffolds. Endochondral ossification is prevented in scaffolds with very small pore sizes; pore interconnectivity is critical to promote capillary ingrowth for mature bone formation. These results provide a novel strategy to control tissue regenerative processes by tunable architecture of macroporous nanofibrous scaffolds.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.