In the era of deepfakes and AI-generated content, digital image manipulation poses significant challenges to authenticity, creating doubts about credibility images. Traditional forensics techniques often struggle detect sophisticated tampering, passive detection approaches are reactive, verifying authenticity only after counterfeiting occurs. this paper, we propose a novel full-resolution secure learned codec (SLIC) designed proactively prevent by self-destructive artifacts upon...

Klebsiella pneumoniae is an important human pathogen causing hospital-acquired and community-acquired infections. Systemic K. infections may be preceded by gastrointestinal colonization, but the basis of this bacterium's interaction with intestinal epithelium remains unclear. Here, we report that Sap (sensitivity to antimicrobial peptides) transporter contributes bacterial-host cell interactions in vivo virulence. Gene deletion showed sapA required for adherence a blood isolate epithelial,...

ABSTRACT Some naturally occurring compounds, known for their antimicrobial activities, have been employed as food additives. However, efficacy in treating infections caused by antibiotic-resistant bacteria is yet to be fully explored. Rapidly growing mycobacteria (RGM), a category within nontuberculous (NTM), are prevalent various environments and can lead humans. The rise of resistance RGM documented concern. In this study, we reported that four specific natural compounds effectively...

End-to-end learned image compression codecs have notably emerged in recent years. These demonstrated superiority over conventional methods, showcasing remarkable flexibility and adaptability across diverse data domains while supporting new distortion losses. Despite challenges such as computational complexity, methods inherently align with learning-based processing analytic pipelines due to their well-suited internal representations. The concept of Video Coding for Machines has garnered...

End-to-end learned image compression codecs have notably emerged in recent years. These demonstrated superiority over conventional methods, showcasing remarkable flexibility and adaptability across diverse data domains while supporting new distortion losses. Despite challenges such as computational complexity, methods inherently align with learning-based processing analytic pipelines due to their well-suited internal representations. The concept of Video Coding for Machines has garnered...

We propose an end-to-end learned image data hiding framework that embeds and extracts secrets in the latent representations of a generic neural compressor. By leveraging perceptual loss function conjunction with our proposed message encoder decoder, approach simultaneously achieves high quality bit accuracy. Compared to existing techniques, offers superior secrecy competitive watermarking robustness compressed domain while accelerating embedding speed by over 50 times. These results...

More than 1,500 fish species are hermaphroditic, but no hermaphroditic lineage appears to be evolutionarily ancient in fishes. Thus, whether more one sex at a time was present during the evolutionary shift from gonochorism hermaphroditism fishes is an intriguing question. Ectopic oocytes were created ovotestes of protandrous black porgy via withdrawal estradiol (E2) administration. These ectopic reprogrammed surrounding cells, which changed Sertoli cells follicle-like cells. We observed that...

We propose an end-to-end learned image compression codec wherein the analysis transform is jointly trained with object classification task. This study affirms that compressed latent representation can predict human perceptual distance judgments accuracy comparable to a custom-tailored DNN-based quality metric. further investigate various neural encoders and demonstrate effectiveness of employing as loss network for tasks beyond judgments. Our experiments show off-the-shelf encoder proves...

The new coronavirus disease 2019 (COVID-19) caused by the severe acute respiratory syndrome (SARS-CoV-2) has been reported and spread globally. There is an urgent need to take measures treat prevent further infection of this virus. Here, we use virtual drug screening establish pharmacophore groups analyze ACE2 binding site spike protein with ZINC database DrugBank molecular docking dynamics simulations. Screening results showed that Venetoclax, a treatment for chronic lymphocytic leukemia,...

Smartphone is the most successful consumer electronic product in today's mobile social network era. The smartphone camera quality and its image post-processing capability dominant factor that impacts consumer's buying decision. However, evaluation of photos taken from smartphones remains a labor-intensive work relies on professional photographers experts. As an extension prior CNN-based NR-IQA approach, we propose multi-task deep CNN model with scene type detection as auxiliary task. With...

In this paper, we propose a music-driven summarization system for home videos based on several content-aware mechanisms. Many audio and video features are employed to help analyzing synchronizing input audios videos. The synchronization is conducted by matching the rhythm of with that audio. Four profiles proposed, which provide users more flexibilities in conducting process. Experiments show good subject test result summarized can be obtained.

Lossy image coding standards such as JPEG and MPEG have successfully achieved high compression rates for human consumption of multimedia data. However, with the increasing prevalence IoT devices, drones, self-driving cars, machines rather than humans are processing a greater portion captured visual content. Consequently, it is crucial to pursue an efficient compressed representation that caters not only vision but also machine tasks. Drawing inspiration from hypothesis in biological systems...

In the era of deepfakes and AI-generated content, digital image manipulation poses significant challenges to authenticity, creating doubts about credibility image. Traditional forensic techniques often struggle detect sophisticated tampering, passive detection approaches are reactive, verifying authenticity only after counterfeiting occurs. this paper, we propose a novel full resolution Secure Learned Image Codec (SLIC) designed proactively prevent by self-destructive artifacts upon...

The digital image manipulation and advancements in Generative AI, such as Deepfake, has raised significant concerns regarding the authenticity of images shared on social media. Traditional forensic techniques, while helpful, are often passive insufficient against sophisticated tampering methods. This paper introduces Secure Learned Image Codec (SLIC), a novel active approach to ensuring through watermark embedding compressed domain. SLIC leverages neural network-based compression embed...

Image steganography is the process of hiding information which can be text, image, or video inside a cover image. The advantage over cryptography that intended secret message does not attract attention and thus more suitable for communication in highly-surveillant environment such as civil disobedience movements. Internet memes social media messaging apps have become popular culture worldwide, so this folk custom good application scenario image steganography. We try to explore adopt...

JPEG has been a widely used lossy image compression codec for nearly three decades. The standard allows to use customized quantization table; however, it's still challenging problem find an optimal table within acceptable computational cost. This work tries solve the dilemma of balancing between cost and specific optimality by introducing new concept texture mosaic images. Instead optimizing single or collection representative images, simulated annealing technique is applied images search...

Artificial Fingerprinting (AF or the so-called digital watermarking) is a technique that can be used to conduct Deepfake attribution by ensuring media authenticity. However, AF does not prioritize its robustness certain kinds of distortions, making embedded watermarks vulnerable some standard image processing operations. Insufficient reduces practicality watermarking techniques. To address this issue, we propose an enhanced distortion agnostic artificial fingerprinting (EDA-AF) framework...

Scalable coding techniques usually focus on enhancing the quality of whole frame when more bandwidth is available. However, not all regions in a are equally important to user perception. This paper proposed user-oriented approach facilitate scalable spatial and temporal domains. By combining attention foveation models, user's (spatial domain) or segment video clip with steady saliency values (temporal prioritized as far perceived concerned.

Lossy image coding standards such as JPEG and MPEG have successfully achieved high compression rates for human consumption of multimedia data. However, with the increasing prevalence IoT devices, drones, self-driving cars, machines rather than humans are processing a greater portion captured visual content. Consequently, it is crucial to pursue an efficient compressed representation that caters not only vision but also machine tasks. Drawing inspiration from hypothesis in biological systems...

We propose an end-to-end learned image data hiding framework that embeds and extracts secrets in the latent representations of a generic neural compressor. By leveraging perceptual loss function conjunction with our proposed message encoder decoder, approach simultaneously achieves high quality bit accuracy. Compared to existing techniques, offers superior secrecy competitive watermarking robustness compressed domain while accelerating embedding speed by over 50 times. These results...