Effect of the Elaboration Method on Structural and Optical Properties of Zn1.33Ga1.335Sn0.33O4:0.5%Cr3+ Persistent Luminescent Nanomaterials.

Near-infrared (NIR) persistent luminescence (PersL) materials have demonstrated promising developments for applications in many advanced fields due to their unique optical properties. Both high-temperature solid-state (SS) or hydrothermal (HT) methods can successfully be used to prepare PersL materials. In this work, Zn1.33Ga1.34Sn0.33O4:0.5%Cr3+ (ZGSO:0.5%Cr3+), a newly proposed nanomaterial for bioimaging, was prepared using SS and HT methods. The results show the crystal structure, morphology and optical properties of the samples that were prepared using both methods. Briefly, the crystallite size of the ZGSO:0.5%Cr3+ prepared using the SS method is ~3 µm, and as expected, is larger than materials prepared using the HT method. However, the growth process used in the hydrothermal environment promotes the formation of ZGSO:0.5%Cr3+ with more uniform shapes and smaller sizes (less than 500 nm). Different diameter ranges of nanoparticles were obtained using HT and ball milling (BM) methods (ranging from 25-50 nm) and by using SS and BM methods (25-200 nm) as well. In addition, the SS-prepared microstructure material has stronger PersL than HT-prepared particles before they go through ball milling to create nanomaterials. On the contrary, after BM treatment, ZGSO:0.5%Cr3+ HT and BM NPs present higher PersL and photoluminescence (PL) properties than ZGSO:0.5%Cr3+ SS and BM NPs, even though both kinds of NPs present worse PersL and PL compared to the original particles before BM. To summarize: preparation methods, whether by SS or HT, with additional grinding as a second step, can have a significant impact on the morphological and luminescent features of ZGSO:0.5%Cr3+ PersL materials.

ZGSO Spinel Nanoparticles with Dual Emission of NIR Persistent Luminescence for Anti-Counterfeiting Applications.

The property of persistent luminescence shows great potential for anti-counterfeiting technology and imaging by taking advantage of a background-free signal. Current anti-counterfeiting technologies face the challenge of low security and the inconvenience of being limited to visible light emission, as emitters in the NIR optical windows are required for such applications. Here, we report the preparation of a series of Zn1+xGa2-2xSnxO4 nanoparticles (ZGSO NPs) with persistent luminescence in the first and second near-infrared window to overcome these challenges. ZGSO NPs, doped with transition-metal (Cr3+ and/or Ni2+) and in some cases co-doped with rare-earth (Er3+) ions, were successfully prepared using an improved solid-state method with a subsequent milling process to reach sub-200 nm size particles. X-ray diffraction and absorption spectroscopy were used for the analysis of the structure and local crystal field around the dopant ions at different Sn4+/Ga3+ ratios. The size of the NPs was ~150 nm, measured by DLS. Doped ZGSO NPs exhibited intense photoluminescence in the range from red, NIR-I to NIR-II, and even NIR-III, under UV radiation, and showed persistent luminescence at 700 nm (NIR-I) and 1300 nm (NIR-II) after excitation removal. Hence, these NPs were evaluated for multi-level anti-counterfeiting technology.

Learning Smooth Representation for Unsupervised Domain Adaptation.

Typical adversarial-training-based unsupervised domain adaptation (UDA) methods are vulnerable when the source and target datasets are highly complex or exhibit a large discrepancy between their data distributions. Recently, several Lipschitz-constraint-based methods have been explored. The satisfaction of Lipschitz continuity guarantees a remarkable performance on a target domain. However, they lack a mathematical analysis of why a Lipschitz constraint is beneficial to UDA and usually perform poorly on large-scale datasets. In this article, we take the principle of utilizing a Lipschitz constraint further by discussing how it affects the error bound of UDA. A connection between them is built, and an illustration of how Lipschitzness reduces the error bound is presented. A local smooth discrepancy is defined to measure the Lipschitzness of a target distribution in a pointwise way. When constructing a deep end-to-end model, to ensure the effectiveness and stability of UDA, three critical factors are considered in our proposed optimization strategy, i.e., the sample amount of a target domain, dimension, and batchsize of samples. Experimental results demonstrate that our model performs well on several standard benchmarks. Our ablation study shows that the sample amount of a target domain, the dimension, and batchsize of samples, indeed, greatly impact Lipschitz-constraint-based methods' ability to handle large-scale datasets. Code is available at https://github.com/CuthbertCai/SRDA.

Conditional Feature Learning Based Transformer for Text-Based Person Search.

Text-based person search aims at retrieving the target person in an image gallery using a descriptive sentence of that person. The core of this task is to calculate a similarity score between the pedestrian image and description, which requires inferring the complex latent correspondence between image sub-regions and textual phrases at different scales. Transformer is an intuitive way to model the complex alignment by its self-attention mechanism. Most previous Transformer-based methods simply concatenate image region features and text features as input and learn a cross-modal representation in a brute force manner. Such weakly supervised learning approaches fail to explicitly build alignment between image region features and text features, causing an inferior feature distribution. In this paper, we present CFLT, Conditional Feature Learning based Transformer. It maps the sub-regions and phrases into a unified latent space and explicitly aligns them by constructing conditional embeddings where the feature of data from one modality is dynamically adjusted based on the data from the other modality. The output of our CFLT is a set of similarity scores for each sub-region or phrase rather than a cross-modal representation. Furthermore, we propose a simple and effective multi-modal re-ranking method named Re-ranking scheme by Visual Conditional Feature (RVCF). Benefit from the visual conditional feature and better feature distribution in our CFLT, the proposed RVCF achieves significant performance improvement. Experimental results show that our CFLT outperforms the state-of-the-art methods by 7.03% in terms of top-1 accuracy and 5.01% in terms of top-5 accuracy on the text-based person search dataset.

Unsupervised Domain Adaptation With Adversarial Residual Transform Networks.

Domain adaptation (DA) is widely used in learning problems lacking labels. Recent studies show that deep adversarial DA models can make markable improvements in performance, which include symmetric and asymmetric architectures. However, the former has poor generalization ability, whereas the latter is very hard to train. In this article, we propose a novel adversarial DA method named adversarial residual transform networks (ARTNs) to improve the generalization ability, which directly transforms the source features into the space of target features. In this model, residual connections are used to share features and adversarial loss is reconstructed, thus making the model more generalized and easier to train. Moreover, a special regularization term is added to the loss function to alleviate a vanishing gradient problem, which enables its training process stable. A series of experiments based on Amazon review data set, digits data sets, and Office-31 image data sets are conducted to show that the proposed ARTN can be comparable with the methods of the state of the art.