CC4S: Encouraging Certainty and Consistency in Scribble-Supervised Semantic Segmentation.

Deep learning-based solutions have achieved impressive performance in semantic segmentation but often require large amounts of training data with fine-grained annotations. To alleviate such requisition, a variety of weakly supervised annotation strategies have been proposed, among which scribble supervision is emerging as a popular one due to its user-friendly annotation way. However, the sparsity and diversity of scribble annotations make it nontrivial to train a network to produce deterministic and consistent predictions directly. To address these issues, in this paper we propose holistic solutions involving the design of network structure, loss and training procedure, named CC4S to improve Certainty and Consistency for Scribble-Supervised Semantic Segmentation. Specifically, to reduce uncertainty, CC4S embeds a random walk module into the network structure to make neural representations uniformly distributed within similar semantic regions, which works together with a soft entropy loss function to force the network to produce deterministic predictions. To encourage consistency, CC4S adopts self-supervision training and imposes the consistency loss on the eigenspace of the probability transition matrix in the random walk module (we named neural eigenspace). Such self-supervision inherits the category-level discriminability from the neural eigenspace and meanwhile helps the network focus on producing consistent predictions for the salient parts and neglect semantically heterogeneous backgrounds. Finally, to further improve the performance, CC4S uses the network predictions as pseudo-labels and retrains the network with an extra color constraint regularizer on pseudo-labels to boost semantic consistency in color space. Rich experiments demonstrate the excellent performance of CC4S. In particular, under scribble supervision, CC4S achieves comparable performance to those from fully supervised methods. Comprehensive ablation experiments verify the effectiveness of the design choices in CC4S and its robustness under extreme supervision cases, i.e., when scribbles are shrunk proportionally or dropped randomly. The code for this work has been open-sourced at https://github.com/panzhiyi/CC4S.

Revealing the closed pore formation of waste wood-derived hard carbon for advanced sodium-ion battery.

Although the closed pore structure plays a key role in contributing low-voltage plateau capacity of hard carbon anode for sodium-ion batteries, the formation mechanism of closed pores is still under debate. Here, we employ waste wood-derived hard carbon as a template to systematically establish the formation mechanisms of closed pores and their effect on sodium storage performance. We find that the high crystallinity cellulose in nature wood decomposes to long-range carbon layers as the wall of closed pore, and the amorphous component can hinder the graphitization of carbon layer and induce the crispation of long-range carbon layers. The optimized sample demonstrates a high reversible capacity of 430 mAh g-1 at 20 mA g-1 (plateau capacity of 293 mAh g-1 for the second cycle), as well as good rate and stable cycling performances (85.4% after 400 cycles at 500 mA g-1). Deep insights into the closed pore formation will greatly forward the rational design of hard carbon anode with high capacity.

Improved Antibacterial Properties of Polylactic Acid-Based Nanofibers Loaded with ZnO-Ag Nanoparticles through Pore Engineering.

In the post-epidemic era, bio-based protective fiber materials with active protective functions are of utmost importance, not only to combat the spread of pathogens but also to reduce the environmental impact of petroleum-based protective materials. Here, efficient antibacterial polylactic acid-based (PLA-based) fibers are prepared by solution blow spinning and their pore structures are regulated by controlling the ratio of the solvent components in the spinning solutions. The porous PLA-based fibers exhibit antibacterial efficiencies of over 99% against Escherichia coli and over 98% against Bacillus subtilis, which are significantly higher than that of the nonporous PLA-based fibers. The excellent antibacterial property of the porous PLA-based fibers can be attributed to their high porosity, which allows antibacterial nanoparticles to be released more easily from the fibers, thus effectively killing pathogenic microorganisms. Moreover, pore structure regulation can also enhance the mechanical property of the PLA-based fiber materials. Our approach of regulating the microstructure and properties of the PLA-based fibers through pore engineering can be extended to other polymer fiber materials and is suitable for polymer-based composite systems that require optimal performance through sufficient exposure of doped materials.

Designing highly efficient 3D porous Ni-Fe sulfide nanosheets based catalyst for the overall water splitting through component regulation.

Constructing efficient and stable bifunctional catalysts is essential to improve the conversion efficiency of overall water splitting (OWS). In this work, 3D porous Ni-Fe sulfide nanosheets supported on nickel foam (Ni-Fe-S/NF) were synthesized by a facile hydrothermal method. The optimized Ni3S2-FeS/NF-2 electrode realized ultra-high efficiency for oxygen evolution reaction (OER) and hydrogen evolution reaction (HER). Benefiting from the unique 3D porous nanosheets structure and the strong electronic interactions between Ni3S2 and FeS through component regulation, low overpotentials of 253 and 262 mV are required to drive a current density of 100 mA cm-2 for OER and HER, respectively. Importantly, Ni3S2-FeS/NF-2 bifunctional catalyst only needs 1.55 and 1.75 V at 10 and 100 mA cm-2 respectively and works continuously for at least 100 h. The work thus provides an extraordinary promising catalyst for OWS and can be envisioned of potential for large-scale applications.

Toward a High-Energy-Density Cathode with Enhanced Temperature Adaptability for Sodium-Ion Batteries: A Case Study of Na3MnZr(PO4)3 Microspheres with Embedded Dual-Carbon Networks.

Polyanionic cathode materials that have high energy density and good temperature adaptability are in high demand for practical applications in sodium-ion batteries (SIBs). In this study, a scalable spray-drying strategy has been proposed to construct interconnected conductive networks composed of amorphous carbon and reduced graphene oxide in Na3MnZr(PO4)3 microspheres (NMZP@C-rGO). The dual-carbon conductive networks provide fast electron migration pathways in the microspheres. Moreover, they significantly increase the porosity and specific surface area of the microspheres, which are conducive to accommodating the volume change and improving the electrode/electrolyte contact interface and the contribution of the pseudocapacitance effect to achieve fast sodium storage. As a result, NMZP@C-rGO exhibits excellent rate performance (50.9 mAh g-1 at 50C and 30 °C, 35.4 mAh g-1 at 50C and -15 °C) and long-term cycling stability (capacity retentions of 97.4 and 79.6% after 1500 cycles at 30 and -15 °C, respectively) in a wide temperature range.

Ultrastrong low-carbon nanosteel produced by heterostructure and interstitial mediated warm rolling.

Ultrastrong materials can notably help with improving the energy efficiency of transportation vehicles by reducing their weight. Grain refinement by severe plastic deformation is, so far, the most effective approach to produce bulk strong nanostructured metals, but its scaling up for industrial production has been a challenge. Here, we report an ultrastrong (2.15 GPa) low-carbon nanosteel processed by heterostructure and interstitial mediated warm rolling. The nanosteel consists of thin (~17.8 nm) lamellae, which was enabled by two unreported mechanisms: (i) improving deformation compatibility of dual-phase heterostructure by adjusting warm rolling temperature and (ii) segregating carbon atoms to lamellar boundaries to stabilize the nanolamellae. Defying our intuition, warm rolling produced finer lamellae than cold rolling, which demonstrates the potential and importance of tuning deformation compatibility of interstitial containing heterostructure for nanocrystallization. This previously unreported approach is applicable to most low-carbon, low-alloy steels for producing ultrahigh strength materials in industrial scale.

Super Diffusion for Salient Object Detection.

One major branch of saliency object detection methods are diffusion-based which construct a graph model on a given image and diffuse seed saliency values to the whole graph by a diffusion matrix. While their performance is sensitive to specific feature spaces and scales used for the diffusion matrix definition, little work has been published to systematically promote the robustness and accuracy of salient object detection under the generic mechanism of diffusion. In this work, we firstly present a novel view of the working mechanism of the diffusion process based on mathematical analysis, which reveals that the diffusion process is actually computing the similarity of nodes with respect to the seeds based on diffusion maps. Following this analysis, we propose super diffusion, a novel inclusive learning-based framework for salient object detection, which makes the optimum and robust performance by integrating a large pool of feature spaces, scales and even features originally computed for non-diffusion-based salient object detection. A closed-form solution of the optimal parameters for the integration is determined through supervised learning. At the local level, we propose to promote each individual diffusion before the integration. Our mathematical analysis reveals the close relationship between saliency diffusion and spectral clustering. Based on this, we propose to re-synthesize each individual diffusion matrix from the most discriminative eigenvectors and the constant eigenvector (for saliency normalization). The proposed framework is implemented and experimented on prevalently used benchmark datasets, consistently leading to state-of-the-art performance.

Microstructure and Mechanical Properties of a Cold-Rolled Ultrafine-Grained Dual-Phase Steel.

A new processing route to produce Ultrafine-Grained Dual-Phase steel has been proposed, involving cold-rolling and subsequent intercritical annealing of a fibrous ferrite⁻martensite starting structure. Ultrafine-grained DP (UFG-DP) steel with an average ferrite grain size of about ~2.7 µm and an average martensite island size of ~2.9 µm was achieved. Tensile testing revealed superior mechanical properties (the ultimate tensile strength of 1267 MPa and uniform elongation of 8.2%) for the new DP steel in comparison with the fibrous DP steels. The superior mechanical properties are attributed to the influence of microstructure refinement on the work-hardening and fracture behavior.