GeoSynth: A Photorealistic Synthetic Indoor Dataset for Scene Understanding.

Deep learning has revolutionized many scene perception tasks over the past decade. Some of these improvements can be attributed to the development of large labeled datasets. The creation of such datasets can be an expensive, time-consuming, and imperfect process. To address these issues, we introduce GeoSynth, a diverse photorealistic synthetic dataset for indoor scene understanding tasks. Each GeoSynth exemplar contains rich labels including segmentation, geometry, camera parameters, surface material, lighting, and more. We demonstrate that supplementing real training data with GeoSynth can significantly improve network performance on perception tasks, like semantic segmentation. A subset of our dataset will be made publicly available at https://github.com/geomagical/GeoSynth.

Adsorption of cellular proteins to polyelectrolyte-functionalized gold nanorods: a mechanism for nanoparticle regulation of cell phenotype?

Cell behavior in the presence of nanomaterials is typically explored through simple viability assays, but there is mounting evidence that nanomaterials can have more subtle effects on a variety of cellular functions. Previously our lab demonstrated that gold nanorods functionalized with polyelectrolyte multi-layers inhibited rat cardiac fibroblast-mediated remodeling of type I collagen scaffolds by altering fibroblast phenotype and the mechanical properties of the collagen network. In this work, we examine a possible mechanism for these effects: adsorption of cellular proteins by the nanorods. Mass spectrometric and gel electrophoresis of media collected from cultured cells suggests that a number of proteins, some of which mediate cell-cell and cell-matrix interactions, adsorb onto the surface of these nanoparticles in vitro. Polyethylene glycol coating of the nanorods largely mitigates protein adsorption and fibroblast-mediated collagen remodeling. These results suggest that adsorption of proteins by nanorods could have a significant effect on cell functions, including fibroblast-mediated matrix remodeling.

High-aspect-ratio gold nanorods: their synthesis and application to image cell-induced strain fields in collagen films.

Gold nanoparticles are receiving considerable attention due to their novel properties and the potential variety of their uses. Long gold nanorods with dimensions of approximately 20 × 400 nm exhibit strong light scattering and can be easily observed under dark-field microscopy. Here we describe the use of this light-scattering property to track micrometer scale strains in collagen gels and thick films, which result from cell traction forces applied by neonatal heart fibroblasts. The use of such collagen constructs to model cell behavior in the extracellular matrix is common, and describing local mechanical environments on such a small scale is necessary to understand the complex factors associated with the remodeling of the collagen network. Unlike other particles used for tracking purposes, gold nanorods do not photobleach, allowing their optical signal to be tracked for longer periods of time, and they can be easily synthesized and coated with various charged or neutral shells, potentially reducing the effect of their presence on the cell system or allowing selective placement. Techniques described here are ultimately applicable for investigations with a wide variety of cells and cell environments.

Formation of PbS nanowire pine trees driven by screw dislocations.

The basic characteristics of nanowire growth driven by screw dislocations were investigated by synthesizing hierarchical lead sulfide (PbS) nanowire "pine trees" using chemical vapor deposition of PbCl(2) and S precursors and systematically observing the effects of various growth parameters, such as hydrogen flow, temperature, pressure, and the growth substrates employed. Statistical surveys showed that the growth rate of the dislocation-driven trunk is about 6 mum/min and that of the vapor-liquid-solid (VLS) driven branch nanowire is about 1.2 mum/min under the typical reaction conditions at 600 degrees C, 900 Torr, and a hydrogen flow rate of 1.5 sccm. The onset of hydrogen flow plus the presence of fresh silicon have been identified as the critical ingredients for generating PbS nanowire trees reproducibly. To explain the experimental findings in the context of classical crystal growth theory, the former is suggested to create a spike in supersaturation of the actual sulfur precursor H(2)S and initiate dislocations with screw components that then propagate anisotropically to form the PbS nanowire trunks. Maintaining suitable hydrogen flow provides a favorable low supersaturation that promotes dislocation-driven trunk nanowire growth and enables the simultaneous VLS nanowire growth of branches. Furthermore, thermodynamic consideration and experiments showed that silicon fortuitously controls the supersaturation by reversibly reacting with H(2)S to form SiS(2) and that SiS(2) can also be a viable precursor for PbS nanowire growth. The key requirements of screw dislocation-driven nanowire growth are summarized. This study provides some general guidelines for further nanowire growth driven by screw dislocations.