Alexandria Williams March 6, 2026Advertisement AI data centers consume vast amounts of water and electricity as servers heat up during use. Engineers are testing new cooling methods — from direct‑to‑chip liquid systems to immersion cooling and even underwater modules in the North Sea. Some propose space‑based data centers, though experts doubt their feasibility. All aim to cut AI’s growing environmental impact.
MagicaL Core is a mobile-first machine learning tool designed for creating, training, and testing image classifiers directly on iPad. Built for accessibility, it removes the need for coding and allows users to work entirely through a visual interface. Users can upload their own photos to train custom models, instantly test classification results, and refine datasets in real time. The workflow is designed for rapid experimentation, making it ideal for learners, designers, and developers exploring machine learning concepts on mobile devices. Once trained, models can be exported for use in external apps or integrated into broader workflows, enabling real-world application beyond experimentation. The platform emphasizes simplicity while still offering meaningful control over training data and model behavior. MagicaL Core turns mobile devices into practical machine learning labs, making AI model creation more accessible and hands-on. Image Classifier Builders MagicaL Core Lets You Train Image Classifiers On iPad Without Code Trend Themes - No-code Model Training — Visual machine learning interfaces are broadening AI creation to non-technical users, opening space for faster prototyping in education, design, and product experimentation. - Mobile AI Labs — Tablet-based development environments turn everyday devices into portable experimentation hubs, reducing dependence on specialized hardware and desktop workflows. - Custom Image Classifiers — User-trained visual recognition models support highly specific use cases, creating new possibilities for niche automation, personalized apps, and domain-focused analytics. Industry Implications - Education Technology — Hands-on AI tools for learners make complex machine learning concepts more accessible, expanding opportunities for interactive curricula and skills-based training. - Mobile Software — App ecosystems gain new value when model creation, testing, and export happen directly on mobile devices, enabling lightweight AI-powered product development. - Computer Vision — Simplified image classification workflows lower barriers to applied visual intelligence, encouraging adoption across specialized business, creative, and operational contexts.
Abstract Multiparametric video-based aquaculture monitoring systems are becoming essential for observing and quantifying fish behavior in real time. These platforms enable high-frequency, continuous recording of underwater activity, supporting automation in feeding management and behavioral analytics. The Tilapia Feeding Behavior Image Dataset (TFBID-mini) was generated using controlled recirculating aquaculture tanks equipped with high-definition underwater cameras to capture feeding events of Oreochromis niloticus (Nile tilapia). The dataset consists of 4,000 expert-annotated images classified into four feeding-intensity levels-None, Weak, Medium, and Strong-each representing a distinct behavioral state observed across multiple feeding sessions. Images were extracted from 1080p video sequences recorded under varying illumination conditions and quality-checked to ensure visual clarity, color balance, and consistency. The resulting dataset provides a standardized benchmarking resource for evaluating computer vision and artificial intelligence models for automated recognition of feeding behavior and for optimizing intelligent feeding systems in aquaculture. Similar content being viewed by others Funding This work is partially supported by the Multi-Source Perception and Intelligent Computing Laboratory of SZTU, the Natural Science Foundation of Top Talent of SZTU (GDRC202411, GDRC202412), the Open Research Fund from Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ) (GML-KF-24-05). Author information Authors and Affiliations Corresponding author Ethics declarations Competing interests The authors declare no competing interests. Additional information Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Rights and permissions Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from
AI is becoming part of the plant world faster than many people expected. Plant identification apps can suggest species from a photo. Greenhouse cameras can help growers spot stress. Flower businesses can organize product images by color, shape, or variety. Researchers can use visual data to track growth, pests, and disease. But AI does not “understand” flowers the way a botanist, grower, or florist does. It learns from examples. It studies thousands or millions of images and looks for patterns. A leaf edge, a petal shape, a stem color, a spot on a leaf, or the structure of a flower head can all become part of what the model learns. The quality of that learning depends on the quality of the data behind it. AI Starts With Images Most plant recognition systems begin with images. These may come from smartphones, greenhouse cameras, drones, field sensors, research collections, or public submissions. An image alone is not enough. The AI needs to know what is in the image. Is it a rose, tulip, monstera, orchid, tomato leaf, or diseased cucumber plant? Is the image showing a flower, leaf, bark, fruit, or whole plant? Is the plant healthy, stressed, mature, young, or damaged? That is where labeling comes in. Behind many plant recognition tools is carefully labeled image data. For businesses building AI models in horticulture, working with a data annotation company can help improve the accuracy of plant, flower, pest, and disease recognition systems. What Data Annotation Means In Plant Recognition Data annotation means adding useful information to an image so AI can learn from it. For a simple plant identification model, the label might be the plant species. For example: sunflower, rose, lavender, peace lily, or Japanese maple. For a more advanced system, the labels may be more detailed. The image
SocietyBrazilBrazil's facial authentication is popular, but risks remainTo view this video please enable JavaScript, and consider upgrading to a web browser that supports HTML5 videoSocietyBrazilNadia Pontes06/25/2026June 25, 2026In Brazil, your face is your ticket. Systems with facial authentication aim to prevent violence and identify suspects. But experts warn about risks to sensitive personal data.https://p.dw.com/p/5G4UsAdvertisement
ShopShield Please wait We are optimizing your request for the best experience.
ShopShield Please wait We are optimizing your request for the best experience.
U.S. Army Lt. Gen. Gregory K. Anderson, commanding general, XVIII Airborne Corps (right), presents Col. Sean R. Shields, outgoing commander, 20th Engineer Brigade, XVIII Airborne Corps (left), with an Order of Saint Michael - Airborne award during a Change of Command Ceremony at Fort Bragg, N.C., June 12, 2026. The award is given in recognition of outstanding contributions to airborne, glider, air assault, and special operations communities. (U.S. Army photo by Cpl. Darius M. Smith) | Date Taken: | 06.12.2026 | | Date Posted: | 06.25.2026 17:52 | | Photo ID: | 9755133 | | VIRIN: | 260612-A-CK855-3377 | | Resolution: | 7520x5016 | | Size: | 6.04 MB | | Location: | FORT BRAGG, NORTH CAROLINA, US | | Web Views: | 9 | | Downloads: | 2 | This work, 20th Engineer Brigade Change of Command [Image 4 of 4], by CPL Darius Smith, identified by DVIDS, must comply with the restrictions shown on https://www.dvidshub.net/about/copyright.
Latest research - Optimizing speech quality and intelligibility through noise-aware enhancement for coding application Uncover the latest and most impactful research in Signal, Speech and Image Processing . Explore pioneering discoveries, insightful ideas and new methods from leading researchers in the field.
Facial recognition clearance extended to Qingmao port and HZMB Today marks the rollout of Smart Immigration Clearance at Qingmao Port and the Zhuhai-Macao Hall of the Hong Kong-Zhuhai-Macao Bridge (HZMB), extending the facial recognition-based system to additional cross-border checkpoints following its earlier launch at Hengqin Port last year. The initiative, which replaces physical document checks with facial recognition technology, was first piloted at Hengqin Port on November 5 last year. The Public Security Police Force (PSP) and the Public Security Forces Affairs Bureau, working alongside the Zhuhai General Station of Exit and Entry Frontier Inspection and the Economic Development Bureau of the Cooperation Zone, upgraded the joint inspection automated channels to allow travelers to simply glance at a scanner rather than use their ID cards. The shift from “card scan” to “face scan” has been widely welcomed by frequent cross-border commuters, who now breeze through checkpoints in seconds. As of June 24, some 310,000 individuals had registered for the service at Hengqin Port, with cumulative usage surpassing 6.21 million passenger trips. That figure represents 42% of all clearances processed through the port’s joint inspection automated channels. Following more than six months of stable operation at Hengqin, authorities decided to extend the service to two additional ports. The Qingmao Port and the HZMB Zhuhai-Macao Hall underwent stress tests from April 22 to 24, followed by trial runs on May 11 and 12. As of June 24, 170,000 travelers had registered at these two locations, and more than 1.72 million trips have been processed using the new system. The smooth trial runs provided authorities with the confidence to proceed with the full-scale rollout. With today’s expansion, all 204 joint inspection automated channels across Macau’s “joint inspection” ports will now support face-scan clearance. The technology simplifies the clearance process significantly, but authorities stress
Antonia Fedder is a digital designer, consultant, and speaker working at the intersection of accessibility, inclusion, and sustainability. She is also a CMS Critic contributor. AI is either the greatest opportunity for the future of digital accessibility or a fast track to cementing existing exclusions. It all depends on who you ask. I honestly think both are happening at the same time, and the decisions that are being made right now will determine where we end up. The disability market remains both chronically underserved and one of the most severely underestimated in the world, and that’s before you consider the much bigger community of people who use the digital world differently: neurodivergent users, senior citizens, people with temporary impairments, anyone who’s ever encountered a digital interface that wasn’t built for them. Getting AI and accessibility right is a market question, a rights question, and a design question. This is not a niche problem. Before we go further, we need to be clear about what we’re actually talking about. "AI" covers a lot of ground. Discriminative AI works through recognition and classification. That’s the technology behind speech recognition systems like Voiceitt, which learns the unique speech patterns of people with cerebral palsy, Parkinson’s, or Down syndrome, so that they can be understood by voice interfaces made only for standard speech. It supports Microsoft’s Seeing AI, which has enabled blind and low-vision users to manage millions of independent tasks using just a smartphone camera. Generative AI creates. Be My AI, built into the Be My Eyes app, goes beyond describing images. It enables actual conversations around them. You can ask follow-up questions and explore what you are seeing. Apple’s Personal Voice allows people living with ALS or other speech-affecting conditions to create a synthetic version of their own voice in 15 minutes
The Tasmanian government has removed more than 468,000 driver’s licence photos from a national facial recognition database, citing concerns about how the biometric data would be protected. State Growth deputy secretary of transport Cynthia Heydon confirmed the recall to a budget estimates committee on June 4. Tasmania uploaded the licence images to a segregated section of the National Driver Licence Facial Recognition Solution between December 2018 and December 2020. The uploads happened without the consent of licence-holders and before federal laws were in place to govern the system. “In July last year, noting we have not entered into a participation agreement with the Australian government, we have chosen to have all of the Tasmanian driver’s licence images removed and associated data removed from that system,” Heydon said. Asked what had happened to the data, Heydon said the government had recovered all of it. The Commonwealth attorney-general’s department confirmed in late 2025 that the Tasmanian data had been removed. Independent Nelson MLC Meg Webb first called for the data to be returned in 2020 and has pushed for it repeatedly since. She welcomed the move but questioned why the government was now mandating similar technology in gambling venues. “If the safety of Tasmanians’ sensitive facial and biometric data cannot be guaranteed when held by a national government database, how on earth can it be considered safe when collected by state gambling venues?” she said. The state government in January shelved a promised pokies pre-commitment card in favour of measures including mandatory facial recognition in venues with gaming machines. “It is not too late. I urge the premier and treasurer to return to the safer and proven effective mandatory pokies pre-commitment card,” Webb said.
Saxony's new police law: AI pursuit and facial recognition approved After months of tug-of-war, the Saxon state parliament has passed far-reaching powers for investigators. Civil rights activists have sharply criticized the move. On Wednesday, after a contentious debate, the Saxon state parliament passed a comprehensive reform of the police law. With a narrow majority of 60 to 53 votes, the parliament paved the way for an extensive expansion of digital surveillance powers in the Free State, with some amendments to the previous government draft. The ruling minority coalition of CDU and SPD was largely dependent on the support of the BSW, from whose ranks eleven of the fifteen members of parliament voted for the legislative proposal. Opposing this were the 'no' votes from the Green, Left, and AfD parliamentary groups. One SPD member abstained. The new regulatory framework, which comes into effect as early as today, Thursday, marks a turning point in Saxon security policy: it equips the police with a broad arsenal of digital investigation tools that reach far into the digital space and public life. Among the core elements of the reform is the permission to use Artificial Intelligence for automated video surveillance at crime hotspots. These systems will in the future be able to track individuals across multiple cameras and recognize behavioral patterns, dangerous objects, and faces in real-time as part of remote biometric identification. Furthermore, the authorities will be empowered to use covert license plate scanners to detect stolen vehicles in flowing traffic. The scope of action is also being expanded on the internet: investigators are allowed to use systems that biometrically match faces and voices with freely accessible network data. For investigations in encrypted messenger services such as WhatsApp, the use of state Trojans for source telecommunications surveillance is also available. In the analog domain,
To ensure we keep this website safe, please can you confirm you are a human by ticking the box below. If you are unable to complete the above request please contact us using the below link, providing a screenshot of your experience. https://ioppublishing.org/contacts/
NJ Supreme Court upholds police FRT disclosure requirement New Jersey Supreme Court has unanimously ruled that law enforcement in the state must disclose its use of facial recognition during investigations as part of the discovery process in court proceedings. The decision partly sides with the state though, in excluding proprietary information like the source code of the facial recognition software used from the discovery demands. The ruling in State v. Miles (via Thomson Reuters) partially reverses the decision by a state appellate court that relied on an interpretation of State v. Arteaga that State Supreme Court Justice Lee Solomon Fasciale described as overly “mechanical” in allowing the demand for source code as part of discovery. The court ruled in Arteaga that police were obligated to disclose 13 pieces of information about the facial recognition system used by detectives before laying charges, which the appellate court took as precedence in Miles. “Such discovery cannot be reduced to a rigid thirteen-item checklist; a defendant’s entitlement varies depending on the specifics of the case,” writes Fasciale. In Miles’ case, New Jersey police must disclose the facial recognition software and data they used before they charged him with first-degree murder, along with how they used it. “Metrics of performance” are included, and Arteaga asked for (and was granted) NIST FRVT results as one of the 13 points, but the key items are the biometric software’s name and developer, what reference database was used, all versions of the probe image used and the match candidates returned. Some of this information was already provided by the prosecution after lower court decisions. Miles’ case had attracted the support of the ACLU, the Innocence Project and the EFF. Article Topics biometrics | facial recognition | false arrest | law enforcement | New Jersey Comments
AI fraud drives banks toward biometric identity defenses New survey data from KPMG shows a majority of banking executives preparing for AI to cause major disruptions to established business models. The 2026 Banking Technology Survey says 80 percent of respondents anticipate “significant shifts” over the next three-to-five years, and are making adjustments accordingly. “What we’re seeing across the banking sector is a convergence of priorities – AI, payments modernization, cybersecurity, and tech-driven M&A are no longer separate agendas,” says Peter Torrente, who leads KPMG’s U.S. Banking Sector activity, in a release. “That shift exposes the challenge to keep pace across technology, risk, and growth simultaneously.” Headline numbers show a cascade effect: 76 percent of banking executives say they have seen an increase in the number of cybersecurity attacks on their banks in the last year. In response, 92 percent are increasing their budgets to address the risk. And 84 percent are increasing cybersecurity investment specifically to address risks introduced by AI. Operational efficiency and automation is receiving the most investment in the next 12 months, but security and fraud prevention is a close second. Among top threats are AI-introduced vulnerabilities in code, deepfakes and securing agentic technologies. Eighty percent of banks are implementing regular audits to ensure data security, privacy and compliance. Notably, uptake in adoption of biometric tools is set to increase; 32 percent of banking leaders say they are already using AI-enabled biometrics to secure payments and access management – but 72 percent plan to integrate the technology over the next three years. Defense tools must evolve at speed of fraud An article from J.P.Morgan dives into fraud in the age of AI. Insights from authors Vincent Meluzio from J.P. Morgan Payments and Douglas Wilbert from Accenture are consistent with the larger industry view: “fraud risk is on
Nestlé is in the midst of a multiyear transformation strategy that has included investing in digital optimizations and creating a "marketing of the future" model. As part of this, in its Ukraine market, the company saw an opportunity to overhaul retail execution that had historically suffered from a lack of real-time visibility in its stores. In 2019, the business segment sought to digitalize its merchandising operations and improve data quality to solve this challenge. The company has worked with effie.ai for the effort, building a digital twin of retail execution so that each store, SKU and action can be captured and verified in near real time. As of this year, the company has completed the following for its 1,500 SKUs across seven categories and 45,000 store visits per month: - Tracked shelf activity via image recognition, GPS and timestamps. - Unified agreements, planograms and promo rules into machine-readable formats. Since implementation, this has increased data accuracy from 75% to 97% while achieving a 60% reduction in shelf audit time, according to effie.ai. Additionally, the company reduced its operational costs and increased promotional availability from 84% to 95%. Nestlé is currently advancing these efforts, aiming to embed agentic functionality.
Misplacing everyday items at home or work is a common frustration that often leads to wasted time and unnecessary stress, but Found App is designed to solve this by helping users build a simple, visual inventory of everything they own. The process is straightforward: users take a photo of each item, add a name or label, and store it within the app. This creates a personal, searchable catalog of belongings that can be accessed instantly whenever needed. Instead of guessing or searching through multiple places, users can quickly locate items using a fast search function. This makes organization more intuitive and reduces the effort required to keep track of important possessions. Found App is built for simplicity, turning everyday organization into a visual and searchable system that works in both home and work environments. Home Inventory Apps Found App Helps You Track And Rediscover Your Belongings Instantly Trend Themes - Visual Inventory Management — Photo-based catalogs are reshaping personal organization by turning everyday belongings into searchable digital records for faster retrieval and reduced household friction. - Searchable Home Systems — Consumer demand for instant item discovery reflects an opening for intelligent home platforms that connect storage, memory, and location data in simpler interfaces. - Personal Asset Digitization — Digital ownership records are expanding beyond insurance and resale use cases into daily convenience tools that make possessions easier to track, manage, and rediscover. Industry Implications - Productivity Apps — Mobile productivity tools are gaining relevance as lightweight inventory features blend task efficiency, visual search, and personal organization into one accessible experience. - Smart Home — Connected-home ecosystems have room for inventory-aware services that link physical spaces with digital catalogs to reduce lost items and improve household visibility. - Workplace Organization — Office and hybrid-work environments benefit from searchable asset tracking that
Offering a variety of advertising and sponsorship options for reaching influential specialists from targeted demographic splits. Cureus provides an equitable, efficient publishing and peer reviewing experience without sacrificing publication times. Generate broad awareness and deliver relevant, peer-reviewed clinical experiences directly to potential customers. Dedicated Cranial Radiosurgery: Clinical Experience with New & Innovative SRS Technologies Sponsored by Zap Surgical Systems Real-Time Adaptive Motion Management on Helical and Robotic RT Platforms Sponsored by Accuray, Inc. Cureus Journal of Medical Science Please note that by doing so you agree to be added to our monthly email newsletter distribution list. By joining Cureus, you agree to our Privacy Policy and Terms of Use.
Figures Abstract Computer-aided diagnosis of skin lesions faces core challenges, including scale diversity, blurred boundaries, intra-class morphological variations, and sparse data. Existing methods often rely on fixed receptive fields or generic attention mechanisms, struggling to fully adapt to the unique characteristics of skin lesions. To address this, we propose an Adaptive Multi-scale Convolutional Attention Network (AMCANet), which aims to achieve accurate and robust classification of skin lesions with limited data. AMCANet comprises three core modules: the adaptive multi-scale convolution module dynamically adjusts the receptive field to accommodate lesions of varying sizes; the hierarchical channel attention module integrates multi-level semantic information across different resolutions; and the skin spatial attention module leverages image gradient information to enhance lesion boundaries and local texture features. Extensive few-shot experiments on the HAM10000 and PAD-UFES-20 public datasets demonstrate that AMCANet significantly outperforms existing baseline models across multiple metrics, exhibiting promising generalization capabilities on the evaluated datasets. Qualitative and visual analyses further validate the model’s ability to extract discriminative features and effectively focus on lesion regions. This study proposes a deep learning model, which demonstrates certain effectiveness in classifying skin lesions even with a small number of samples, providing a potential direction for future research. Citation: Jin H, Liu E, Xu Q, Li Y, Chen X, Chen W (2026) Few-shot skin lesion classification with Adaptive Multi-Scale Convolutional Attention Network. PLoS One 21(6): e0351318. https://doi.org/10.1371/journal.pone.0351318 Editor: Amgad Muneer, The University of Texas, MD Anderson Cancer Center, UNITED STATES OF AMERICA Received: December 16, 2025; Accepted: May 26, 2026; Published: June 25, 2026 Copyright: © 2026 Jin et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: The data are provided