Rendering synthetic data (e.g., 3D CAD-rendered images) to generate annotations for learning deep models in vision tasks has attracted increasing attention in recent years. However, simply applying the models learnt on synthetic images may lead to high generalization error on real images due to domain shift. To address this issue, recent progress in cross-domain recognition has featured the Mean Teacher, which directly simulates unsupervised domain adaptation as semi-supervised learning. The domain gap is thus naturally bridged with consistency regularization in a teacher-student scheme. In this work, we advance this Mean Teacher paradigm to be applicable for crossdomain detection. Specifically, we present Mean Teacher with Object Relations (MTOR) that novelly remolds Mean Teacher under the backbone of Faster R-CNN by integrating the object relations into the measure of consistency cost between teacher and student modules. Technically, MTOR firstly learns relational graphs that capture similarities between pairs of regions for teacher and student respectively. The whole architecture is then optimized with three consistency regularizations: 1) region-level consistency to align the region-level predictions between teacher and student, 2) inter-graph consistency for matching the graph structures between teacher and student, and 3) intra-graph consistency to enhance the similarity between regions of same class within the graph of student. Extensive experiments are conducted on the transfers across Cityscapes, Foggy Cityscapes, and SIM10k, and superior results are reported when comparing to state-of-the-art approaches. More remarkably, we obtain a new record of single model: 22.8% of mAP on Syn2Real detection dataset.
The transport and the delivery of drugs through nanocarriers is a great challenge of pharmacology. Since the production of liposomes to reduce the toxicity of doxorubicin in patients, a plethora of nanomaterials have been produced and characterized. Although it is widely known that elementary properties of nanomaterials influence their in vivo kinetics, such interaction is often poorly investigated in many preclinical studies. The present study aims to evaluate the actual effect of size and shape on the biodistribution of a set of gold nanoparticles (GNPs) after intravenous administration in mice. To this goal, quantitative data achieved by inductively coupled plasma mass spectrometry and observational results emerging from histochemistry (autometallography and enhanced dark-field hyperspectral microscopy) were combined. Since the immune system plays a role in bionano-interaction we used healthy immune-competent mice. To keep the immune surveillance on the physiological levels we synthesized endotoxin-free GNPs to be tested in specific pathogen-free animals. Our study mainly reveals that (a) the size and the shape greatly influence the kinetics of accumulation and excretion of GNPs in filter organs; (b) spherical and star-like GNPs showed the same percentage of accumulation, but a different localization in liver; (c) only star-like GNPs are able to accumulate in lung; (d) changes in the geometry did not improve the passage of the blood brain barrier. Overall, this study can be considered as a reliable starting point to drive the synthesis and the functionalization of potential candidates for theranostic purposes in many fields of research.
In this paper, we introduce the new ideas of augmenting Convolutional Neural Networks (CNNs) with Memory and learning to learn the network parameters for the unlabelled images on the fly in one-shot learning. Specifically, we present Memory Matching Networks (MM-Net) -a novel deep architecture that explores the training procedure, following the philosophy that training and test conditions must match. Technically, MM-Net writes the features of a set of labelled images (support set) into memory and reads from memory when performing inference to holistically leverage the knowledge in the set. Meanwhile, a Contextual Learner employs the memory slots in a sequential manner to predict the parameters of CNNs for unlabelled images. The whole architecture is trained by once showing only a few examples per class and switching the learning from minibatch to minibatch, which is tailored for one-shot learning when presented with a few examples of new categories at test time. Unlike the conventional one-shot learning approaches, our MM-Net could output one unified model irrespective of the number of shots and categories. Extensive experiments are conducted on two public datasets, i.e., Omniglot and miniImageNet, and superior results are reported when compared to state-of-the-art approaches. More remarkably, our MM-Net improves one-shot accuracy on Omniglot from 98.95% to 99.28% and from 49.21% to 53.37% on miniImageNet.
Everywhere in our surroundings we increasingly come in contact with nanostructures that have distinctive complex shape features on a scale comparable to the particle itself. Such shape ensembles can be made by modern nano-synthetic methods and many industrial processes. With the ever growing universe of nanoscale shapes, names such as "nanoflowers" and "nanostars" no longer precisely describe or characterise the distinct nature of the particles. Here we capture and digitise particle shape information on the relevant size scale and create a condensed representation in which the essential shape features can be captured, recognized and correlated. We find the natural emergence of intrinsic shape groups as well-defined ensemble distributions and show how these may be analyzed and interpreted to reveal novel aspects of our nanoscale shape environment. We show how these ideas may be applied to the interaction between the nanoscale-shape and the living universe and provide a conceptual framework for the study of nanoscale shape biological recognition and identity.
The blood−brain barrier (BBB) is highly selective and acts as the interface between the central nervous system and circulation. While the BBB is critical for maintaining brain homeostasis, it represents a formidable challenge for drug delivery.Here we synthesized gold nanoparticles (AuNPs) for targeting the tight junction specifically and demonstrated that transcranial picosecond laser stimulation of these AuNPs post intravenous injection increases the BBB permeability. The BBB permeability change can be graded by laser intensity, is entirely reversible, and involves increased paracellular diffusion. BBB modulation does not lead to significant disruption in the spontaneous vasomotion or the structure of the neurovascular unit. This strategy allows the entry of immunoglobulins and viral gene therapy vectors, as well as cargoladen liposomes. We anticipate this nanotechnology to be useful for tissue regions that are accessible to light or fiberoptic application and to open new avenues for drug screening and therapeutic interventions in the central nervous system.
Triple-localized WHIRLY2 plays a vital role in carbon reallocation from maternal tissue to filial tissue through controlling sucrose transporter gene expression.
The range of possible nanostructures is so large and continuously growing, that collating and unifying the knowledge connected to them, including their biological activity, is a major challenge.
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.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.