Conditional Boundary Loss for Semantic Segmentation.

Improving boundary segmentation results has recently attracted increasing attention in the field of semantic segmentation. Since existing popular methods usually exploit the long-range context, the boundary cues are obscure in the feature space, leading to poor boundary results. In this paper, we propose a novel conditional boundary loss (CBL) for semantic segmentation to improve the performance of the boundaries. The CBL creates a unique optimization goal for each boundary pixel, conditioned on its surrounding neighbors. The conditional optimization of the CBL is easy yet effective. In contrast, most previous boundary-aware methods have difficult optimization goals or may cause potential conflicts with the semantic segmentation task. Specifically, the CBL enhances the intra-class consistency and inter-class difference, by pulling each boundary pixel closer to its unique local class center and pushing it away from its different-class neighbors. Moreover, the CBL filters out noisy and incorrect information to obtain precise boundaries, since only surrounding neighbors that are correctly classified participate in the loss calculation. Our loss is a plug-and-play solution that can be used to improve the boundary segmentation performance of any semantic segmentation network. We conduct extensive experiments on ADE20K, Cityscapes, and Pascal Context, and the results show that applying the CBL to various popular segmentation networks can significantly improve the mIoU and boundary F-score performance.

Pose-Guided Hierarchical Semantic Decomposition and Composition for Human Parsing.

Human parsing is a fine-grained semantic segmentation task, which needs to understand human semantic parts. Most existing methods model human parsing as a general semantic segmentation, which ignores the inherent relationship among hierarchical human parts. In this work, we propose a pose-guided hierarchical semantic decomposition and composition framework for human parsing. Specifically, our method includes a semantic maintained decomposition and composition (SMDC) module and a pose distillation (PC) module. SMDC progressively disassembles the human body to focus on the more concise regions of interest in the decomposition stage and then gradually assembles human parts under the guidance of pose information in the composition stage. Notably, SMDC maintains the atomic semantic labels during both stages to avoid the error propagation issue of the hierarchical structure. To further take advantage of the relationship of human parts, we introduce pose information as explicit guidance for the composition. However, the discrete structure prediction in pose estimation is against the requirement of the continuous region in human parsing. To this end, we design a PC module to broadcast the maximum responses of pose estimation to form the continuous structure in the way of knowledge distillation. The experimental results on the look-into-person (LIP) and PASCAL-Person-Part datasets demonstrate the superiority of our method compared with the state-of-the-art methods, that is, 55.21% mean Intersection of Union (mIoU) on LIP and 69.88% mIoU on PASCAL-Person-Part.

Polymers zippered-up by electric charge reveal themselves.

In the current issue of ACS Nano, Löbling, Haataja et al. craft polymeric nanoparticles with a hierarchy of nontrivial surface structures by combining conventional interpolyelectrolyte complexation with steric control from an uncharged copolymer block. Remarkable cylindrical and lamellar nanodomains are produced on the polyionic coronae of spherical micelles. Here, we discuss generalizing this elegant self-assembly strategy and provide speculative perspectives for its future potential for new nanomaterials.

Revisiting polymer surface diffusion in the extreme case of strong adsorption.

Revisiting polymer surface adsorption with a level of quantification not possible at the time of earlier seminal contributions to this field, we employ fluorescence microscopy to quantify the in-plane diffusion of end-labeled polystyrene adsorbed onto quartz and mica from cyclohexane solution, mostly at 25 °C. Care is taken to prohibit a surface-hopping mechanism, and the experimental techniques are adapted to measurements that persist for up to a few days. The main conclusion is that we fail to observe a single Fickian diffusion coefficient: instead, diffusion displays a broad multicomponent spectrum, indicating that the heterogeneity of surface diffusion fails to average out even over these long times and over distances (∼600 nm, the diameter of a diffraction-limited spot) greatly exceeding the size of the polymer molecules. This holds generally when we vary the molecular weight, the surface roughness, and the temperature. It quantifies the long-believed scenario that strongly adsorbed polymer layers (monomer-surface interaction of more than 1k(B)T) intrinsically present diverse surface conformations that present heterogeneous environments to one another as they diffuse. Bearing in mind that in spite of adsorption from dilute solution the interfacial polymer concentration is high, ramifications of these findings are relevant to the interfacial mobility of polymer glasses, melts, and nanocomposites.

Super-resolution study of polymer mobility fluctuations near c*.

Nanoscale dynamic heterogeneities in synthetic polymer solutions are detected using super-resolution optical microscopy. To this end, we map concentration fluctuations in polystyrene-toluene solutions with spatial resolution below the diffraction limit, focusing on critical fluctuations near the polymer overlap concentration, c*. Two-photon super-resolution microscopy was adapted to be applicable in an organic solvent, and a home-built STED-FCS system with stimulated emission depletion (STED) was used to perform fluorescence correlation spectroscopy (FCS). The polystyrene serving as the tracer probe (670 kg mol(-1), radius of gyration RG ≈ 35 nm, end-labeled with a bodipy derivative chromophore) was dissolved in toluene at room temperature (good solvent) and mixed with matrix polystyrene (3,840 kg mol(-1), RG ≈ 97 nm, Mw/Mn = 1.04) whose concentration was varied from dilute to more than 10c*. Whereas for dilute solutions the intensity-intensity correlation function follows a single diffusion process, it splits starting at c* to imply an additional relaxation process provided that the experimental focal area does not greatly exceed the polymer blob size. We identify the slower mode as self-diffusion and the increasingly rapid mode as correlated segment fluctuations that reflect the cooperative diffusion coefficient, Dcoop. These real-space measurements find quantitative agreement between correlation lengths inferred from dynamic measurements and those from determining the limit below which diffusion coefficients are independent of spot size. This study is considered to illustrate the potential of importing into polymer science the techniques of super-resolution imaging.

Selective Janus particle assembly at tipping points of thermally-switched wetting.

Thermal wetting can simply, selectively and reversibly join patchy particles into clusters (2D and 3D) and also colloidal crystals over the narrow temperature range of 1-2 °C. This is demonstrated with Janus particles (gold half-coated silica spheres) immersed in a binary mixture of water/2,6-lutidine, such that the relative strength of gold-gold bonding through hydrophobic interaction and silica-silica bonding through the wetting-induced attraction is reversibly switched according to temperature.

Single-molecule observation of long jumps in polymer adsorption.

Single-molecule fluorescence imaging of adsorption onto initially bare surfaces shows that polymer chains need not localize immediately after arrival. In a system optimized to present limited adsorption sites (quartz surface to which polyethylene glycol (PEG) chains adsorb from aqueous solution at pH 8.2), we find that some chains diffuse back into bulk solution and readsorb at some distance away, sometimes multiple times before they either localize at a stable position or diffuse away into bulk solution. This mechanism of surface diffusion is considerably more rapid than the classical model in which adsorbed polymers crawl on surfaces while the entire molecule remains adsorbed, suggesting the conceptual generality of a recent report ( Phys. Rev. Lett. 2013 , 110 , 256101 ) but in a new experimental system and with comparison of different chain lengths. We find the trajectories with jumps to follow a truncated Lévy distribution of step size with limiting slope -2.5, consistent with a well-defined, rapid surface diffusion coefficient over the times we observe. The broad waiting time distribution appears to reflect that polymer chains possess a broad distribution of bound fraction: the smaller the bound fraction of a given chain, the shorter the surface residence time before executing the next surface jump.

Effects of anions on the aggregation of charged microgels.

In the present study, we have prepared poly(N-isopropylacrylamide-co-vinylbenzyl trimethylammonium chloride) (NIPAM-co-VT) and poly(N-isopropylacrylamide-co-sodium acrylate) (NIPAM-co-SA) thermally sensitive microgels, which are positively and negatively charged, respectively. By use of laser light scattering (LLS), we have investigated the temperature-induced aggregation of microgels in the presence of salts with the same cation but different anions. The studies demonstrate that neither kosmotropic nor chaotropic anions can lead to the aggregation of NIPAM-co-VT microgels. No obvious specific anion effects can be observed in the phase transition of NIPAM-co-VT microgels. However, kosmotropic and chaotropic anions lead to the aggregation of NIPAM-co-SA microgels. The ordering of anions based on the aggregation temperature at the same salt concentration is consistent with the typical Hofmeister series. The aggregation is thought to be determined by the indirect interfacial effects of anions near the microgel surface. Additionally, the aggregate size of microgels increases with salt concentration for the kosmotropic anions.

Chemical oscillation induced periodic swelling and shrinking of a polymeric multilayer investigated with a quartz crystal microbalance.

Poly(acrylic acid- co-3-azidopropyl acrylate) and poly(acrylic acid- co-propargyl acrylate) have been alternately fabricated into a multilayer via the click reaction. The layer-by-layer deposition was monitored with a quartz crystal microbalance with dissipation (QCM-D) in real time. The response of the multilayer under continuous flow of a bromate-sulfite-ferrocyanide solution with pH oscillation has also been investigated by use of QCM-D. As the pH oscillates between 3.1 and 6.6, either the frequency shift (Delta f) or the dissipation shift (Delta D) periodically varies with a constant amplitude, clearly indicating that the multilayer swells and shrinks oscillatedly. The changes of thickness, shear viscosity, and elastic shear modulus further indicate the oscillation.