This paper introduces a method for quantifying the transferability of architectural components in Single Image Super-Resolution (SISR) models, termed "Universality," and proposes a Universality Assessment Equation (UAE). Guided by the UAE, the authors design optimized modules, Cycle Residual Block (CRB) and Depth-Wise Cycle Residual Block (DCRB), and demonstrate their effectiveness across various datasets and low-level tasks. Results show that networks using these modules outperform state-of-the-art methods, achieving improved PSNR or parameter reduction.
KAUST researchers have published a study in Nature Genetics detailing genomic analysis of wild rice relatives. The study examined nine tetraploid and two diploid wild relatives of rice, finding significant genetic diversity due to transposable elements. This diversity includes genes that confer resilience to heat, drought, and salinity. Why it matters: These findings can help improve rice yields, introduce rice cultivation to currently untenable regions, and protect rice crops against climate change, especially in the Middle East.
The paper introduces MedNNS, a neural network search framework designed for medical imaging, addressing challenges in architecture selection and weight initialization. MedNNS constructs a meta-space encoding datasets and models based on their performance using a Supernetwork-based approach, expanding the model zoo size by 51x. The framework incorporates rank loss and Fréchet Inception Distance (FID) loss to capture inter-model and inter-dataset relationships, improving alignment in the meta-space and outperforming ImageNet pre-trained DL models and SOTA NAS methods.
This paper introduces SemDiff, a novel method for generating unrestricted adversarial examples (UAEs) by exploring the semantic latent space of diffusion models. SemDiff uses multi-attribute optimization to ensure attack success while preserving the naturalness and imperceptibility of generated UAEs. Experiments on high-resolution datasets demonstrate SemDiff's superior performance compared to state-of-the-art methods in attack success rate and imperceptibility, while also evading defenses.
Researchers from MBZUAI introduced RP-SAM2, a method to improve surgical instrument segmentation by refining point prompts for more stable results. RP-SAM2 uses a novel shift block and compound loss function to reduce sensitivity to point prompt placement, improving segmentation accuracy in data-constrained settings. Experiments on the Cataract1k and CaDIS datasets show that RP-SAM2 enhances segmentation accuracy and reduces variance compared to SAM2, with code available on GitHub.
KAUST researchers have captured the initial unwinding of DNA using cryo-electron microscopy and deep learning. The study details 15 atomic states describing how the Simian Virus 40 Large Tumor Antigen helicase unwinds DNA, revealing the coordinated roles of DNA, helicases, and ATP. The research elucidates the fundamental mechanisms of DNA replication, a cornerstone of growth and reproduction. Why it matters: This detailed understanding of helicase function could lead to advances in nanotechnology and our understanding of genetic processes.
Researchers introduce SALT, a parameter-efficient fine-tuning method for medical image segmentation that combines singular value adaptation with low-rank transformation. SALT selectively adapts influential singular values and complements this with a low-rank update for the remaining subspace. Experiments on five medical datasets show SALT outperforms state-of-the-art PEFT methods by 2-5% in Dice score with only 3.9% trainable parameters.
Researchers at MBZUAI have introduced a novel approach to enhance Large Multimodal Models (LMMs) for autonomous driving by integrating 3D tracking information. This method uses a track encoder to embed spatial and temporal data, enriching visual queries and improving the LMM's understanding of driving scenarios. Experiments on DriveLM-nuScenes and DriveLM-CARLA benchmarks demonstrate significant improvements in perception, planning, and prediction tasks compared to baseline models.
Researchers introduce TimeTravel, a benchmark dataset for evaluating large multimodal models (LMMs) on historical and cultural artifacts. The benchmark comprises 10,250 expert-verified samples across 266 cultures and 10 historical regions, designed to assess AI in tasks like classification and interpretation of manuscripts, artworks, inscriptions, and archaeological discoveries. The goal is to establish AI as a reliable partner in preserving cultural heritage and assisting researchers.
This paper introduces Adaptive Entropy-aware Optimization (AEO), a new framework to tackle Multimodal Open-set Test-time Adaptation (MM-OSTTA). AEO uses Unknown-aware Adaptive Entropy Optimization (UAE) and Adaptive Modality Prediction Discrepancy Optimization (AMP) to distinguish unknown class samples during online adaptation by amplifying the entropy difference between known and unknown samples. The study establishes a new benchmark derived from existing datasets with five modalities and evaluates AEO's performance across various domain shift scenarios, demonstrating its effectiveness in long-term and continual MM-OSTTA settings.
The paper introduces VENOM, a text-driven framework for generating high-quality unrestricted adversarial examples using diffusion models. VENOM unifies image content generation and adversarial synthesis into a single reverse diffusion process, enhancing both attack success rate and image quality. The framework incorporates an adaptive adversarial guidance strategy with momentum to ensure the generated adversarial examples align with the distribution of natural images.
MBZUAI researchers introduce UniMed-CLIP, a unified Vision-Language Model (VLM) for diverse medical imaging modalities, trained on the new large-scale, open-source UniMed dataset. UniMed comprises over 5.3 million image-text pairs across six modalities: X-ray, CT, MRI, Ultrasound, Pathology, and Fundus, created using LLMs to transform classification datasets into image-text formats. UniMed-CLIP significantly outperforms existing generalist VLMs and matches modality-specific medical VLMs in zero-shot evaluations, improving over BiomedCLIP by +12.61 on average across 21 datasets while using 3x less training data.
MBZUAI releases BiMediX2, a bilingual (Arabic-English) Bio-Medical Large Multimodal Model, along with the BiMed-V dataset (1.6M samples) and BiMed-MBench evaluation benchmark. BiMediX2 supports multi-turn conversation in Arabic and English and handles diverse medical imaging modalities. The model achieves state-of-the-art results on medical LLM and LMM benchmarks, outperforming existing methods and GPT-4 in specific evaluations.
A new study uses the UNet++ deep learning model and Sentinel-2 satellite data to monitor mangrove dynamics in the UAE from 2017 to 2024. The model achieved a mean Intersection over Union (mIoU) of 87.8% on the validation set. Results indicate a significant increase in mangrove area, primarily in Abu Dhabi, contributing to enhanced carbon sequestration across the UAE.
KAUST researchers have discovered biological clues in the Wahbah Crater in Saudi Arabia that could provide insights into the possibility of life on Enceladus, one of Saturn's moons. The researchers isolated 48 bacterial strains from the crater, identifying two with an adaptability suitable for the extreme environment of Enceladus. These strains thrive in high temperatures, salinity, and alkaline pH levels, mimicking conditions on the Saturn moon. Why it matters: This study highlights the potential of Saudi Arabia's extreme environments as valuable models for detecting extraterrestrial life and strengthens the country's growing interest in space exploration.
FancyVideo, a new video generator, introduces a Cross-frame Textual Guidance Module (CTGM) to enhance text-to-video models. CTGM uses a Temporal Information Injector and Temporal Affinity Refiner to achieve frame-specific textual guidance, improving comprehension of temporal logic. Experiments on the EvalCrafter benchmark demonstrate FancyVideo's state-of-the-art performance in generating dynamic and consistent videos, also supporting image-to-video tasks.
MBZUAI researchers introduce Web2Code, a new large-scale dataset and evaluation framework for training and benchmarking multimodal LLMs on webpage understanding and HTML code generation. The dataset includes webpage images, HTML code, and QA pairs about webpage content. Experiments demonstrate the dataset's utility in webpage understanding, code generation, and general visual domain tasks, with code and data available on Github.
MBZUAI researchers introduce VideoGPT+, a novel video Large Multimodal Model (LMM) that integrates image and video encoders to leverage both spatial and temporal information in videos. They also introduce VCGBench-Diverse, a comprehensive benchmark for evaluating video LMMs across 18 video categories. VideoGPT+ demonstrates improved performance on multiple video benchmarks, including VCGBench and MVBench.
This paper introduces a deep vision-based framework for predicting coastal floods under climate change, addressing the challenges of limited training data and high-dimensional output. The framework employs and compares various deep learning models, including a custom compact CNN architecture, against geostatistical and traditional machine learning methods. A new synthetic dataset of flood inundation maps for Abu Dhabi's coast is also provided to benchmark future models.
This survey paper reviews recent literature on continual learning in medical imaging, addressing challenges like catastrophic forgetting and distribution shifts. It covers classification, segmentation, detection, and other tasks, while providing a taxonomy of studies and identifying challenges. The authors also maintain a GitHub repository to keep the survey up-to-date with the latest research.
Researchers from MBZUAI have introduced the Complex Video Reasoning and Robustness Evaluation Suite (CVRR-ES) for assessing Video-LLMs. The benchmark evaluates models across 11 real-world video dimensions, revealing challenges in robustness and reasoning, particularly for open-source models. A training-free Dual-Step Contextual Prompting (DSCP) technique is proposed to enhance Video-LMM performance, with the dataset and code made publicly available.
Researchers at MBZUAI have developed DynaMMo, a dynamic model merging method for efficient class incremental learning using medical images. DynaMMo merges multiple networks at different training stages using lightweight learnable modules, reducing computational overhead. Evaluated on three datasets, DynaMMo achieved a 10-fold reduction in GFLOPS compared to existing dynamic methods with a 2.76 average accuracy drop.
The paper introduces MedPromptX, a clinical decision support system using multimodal large language models (MLLMs), few-shot prompting (FP), and visual grounding (VG) for chest X-ray diagnosis, integrating imagery with EHR data. MedPromptX refines few-shot data dynamically for real-time adjustment to new patient scenarios and narrows the search area in X-ray images. The study introduces MedPromptX-VQA, a new visual question answering dataset, and demonstrates state-of-the-art performance with an 11% improvement in F1-score compared to baselines.
Researchers at MBZUAI introduce FissionFusion, a hierarchical model merging approach to improve medical image analysis performance. The method uses local and global aggregation of models based on hyperparameter configurations, along with a cyclical learning rate scheduler for efficient model generation. Experiments show FissionFusion outperforms standard model souping by approximately 6% on HAM10000 and CheXpert datasets and improves OOD performance.
Researchers at MBZUAI have introduced TiBiX, a novel approach leveraging temporal information from previous chest X-rays (CXRs) and reports for bidirectional generation of current CXRs and reports. TiBiX addresses two key challenges: generating current images from previous images and reports, and generating current reports from both previous and current images. The study also introduces a curated temporal benchmark dataset derived from the MIMIC-CXR dataset and achieves state-of-the-art results in report generation.
Researchers at MBZUAI have introduced MedMerge, a transfer learning technique that merges weights from independently initialized models to improve performance on medical imaging tasks. MedMerge learns kernel-level weights to combine features from different models into a single model. Experiments across various medical imaging tasks demonstrated performance gains of up to 7% in F1 score.
This paper introduces a new Single Domain Generalization (SDG) method called ConDiSR for medical image classification, using channel-wise contrastive disentanglement and reconstruction-based style regularization. The method is evaluated on multicenter histopathology image classification, achieving a 1% improvement in average accuracy compared to state-of-the-art SDG baselines. Code is available at https://github.com/BioMedIA-MBZUAI/ConDiSR.
Researchers from MBZUAI have developed XReal, a diffusion model for generating realistic chest X-ray images with precise control over anatomy and pathology location. The model utilizes an Anatomy Controller and a Pathology Controller to introduce spatial control in a pre-trained Text-to-Image Diffusion Model without fine-tuning. XReal outperforms existing X-ray diffusion models in realism, as evaluated by quantitative metrics and radiologists' ratings, and the code/weights are available.
Researchers introduce PALO, a polyglot large multimodal model with visual reasoning capabilities in 10 major languages including Arabic. A semi-automated translation approach was used to adapt the multimodal instruction dataset from English to the target languages. The models are trained across three scales (1.7B, 7B and 13B parameters) and a multilingual multimodal benchmark is proposed for evaluation.
Researchers introduce TomFormer, a transformer-based model for accurate and early detection of tomato leaf diseases, with the goal of deployment on the Hello Stretch robot for real-time diagnosis. TomFormer combines a visual transformer and CNN, achieving state-of-the-art results on KUTomaDATA, PlantDoc, and PlantVillage datasets. KUTomaDATA was collected from a greenhouse in Abu Dhabi, UAE.
Researchers propose a universal anatomical embedding (UAE) framework for medical image analysis to learn appearance, semantic, and cross-modality anatomical embeddings. UAE incorporates semantic embedding learning with prototypical contrastive loss, a fixed-point-based matching strategy, and an iterative approach for cross-modality embedding learning. The framework was evaluated on landmark detection, lesion tracking and CT-MRI registration tasks, outperforming existing state-of-the-art methods.
Researchers at MBZUAI have developed GeoChat, a new vision-language model (VLM) specifically designed for remote sensing imagery. GeoChat addresses the limitations of general-domain VLMs in accurately interpreting high-resolution remote sensing data, offering both image-level and region-specific dialogue capabilities. The model is trained on a novel remote sensing multimodal instruction-following dataset and demonstrates strong zero-shot performance across tasks like image captioning and visual question answering.
MBZUAI researchers introduce PG-Video-LLaVA, a large multimodal model with pixel-level grounding capabilities for videos, integrating audio cues for enhanced understanding. The model uses an off-the-shelf tracker and grounding module to localize objects in videos based on user prompts. PG-Video-LLaVA is evaluated on video question-answering and grounding benchmarks, using Vicuna instead of GPT-3.5 for reproducibility.
Researchers at the Technology Innovation Institute (TII) have released a fully-annotated dataset for autonomous drone racing, called "Race Against the Machine." The dataset includes high-resolution visual, inertial, and motion capture data from both autonomous and piloted flights, along with commands, control inputs, and corner-level labeling of drone racing gates. The specifications to recreate their flight platform using commercial off-the-shelf components and the Betaflight controller are also released. Why it matters: This comprehensive resource aims to support the development of new methods and establish quantitative comparisons for approaches in robotics and AI, democratizing drone racing research.
This paper introduces a domain generalization (DG) method for Diabetic Retinopathy (DR) classification that maximizes mutual information using a large pretrained model. The method aims to address the challenge of domain shift in medical imaging caused by variations in data acquisition. Experiments on public datasets demonstrate that the proposed method outperforms state-of-the-art techniques, achieving a 5.25% improvement in average accuracy.
This paper introduces Pulmonary Embolism Detection using Contrastive Learning (PECon), a supervised contrastive pretraining strategy using both CT scans and EHR data to improve feature alignment between modalities for better PE diagnosis. PECon pulls sample features of the same class together while pushing away features of other classes. The approach achieves state-of-the-art results on the RadFusion dataset, with an F1-score of 0.913 and AUROC of 0.943.
This paper introduces a convolutional transformer model for classifying tomato maturity, along with a new UAE-sourced dataset, KUTomaData, for training segmentation and classification models. The model combines CNNs and transformers and was tested against two public datasets. Results showed state-of-the-art performance, outperforming existing methods by significant margins in mAP scores across all three datasets.
MBZUAI researchers introduce XrayGPT, a conversational medical vision-language model for analyzing chest radiographs and answering open-ended questions. The model aligns a medical visual encoder (MedClip) with a fine-tuned large language model (Vicuna) using a linear transformation. To enhance performance, the LLM was fine-tuned using 217k interactive summaries generated from radiology reports.
Video-ChatGPT is a new multimodal model that combines a video-adapted visual encoder with a large language model (LLM) to enable detailed video understanding and conversation. The authors introduce a new dataset of 100,000 video-instruction pairs for training the model. They also develop a quantitative evaluation framework for video-based dialogue models.
This paper introduces Provable Unrestricted Adversarial Training (PUAT), a novel adversarial training approach. PUAT enhances robustness against both unrestricted and restricted adversarial examples while improving standard generalizability by aligning the distributions of adversarial examples, natural data, and the classifier's learned distribution. The approach uses partially labeled data and an augmented triple-GAN to generate effective unrestricted adversarial examples, demonstrating superior performance on benchmarks.
Researchers from KAUST and Beneath The Waves used camera tags on tiger sharks in the Bahamas to map the world's largest seagrass ecosystem, estimated at 92,000 km2. The study, published in Nature Communications, details how the sharks' movements helped validate spatial estimates of these blue carbon ecosystems. Carlos Duarte (KAUST) realized the potential of the shark movement data to map the extensive seagrass meadows in the shallow sand banks of the Bahamas. Why it matters: The innovative approach demonstrates a novel method for mapping and quantifying vast underwater ecosystems, crucial for understanding and conserving blue carbon sinks.
Researchers from MBZUAI have developed EchoCoTr, a novel spatiotemporal deep learning method for estimating left ventricular ejection fraction (LVEF) from echocardiograms. EchoCoTr combines CNNs and vision transformers to overcome the limitations of each when applied to medical video data. The method achieves state-of-the-art results on the EchoNet-Dynamic dataset, demonstrating improved accuracy compared to existing approaches, with code available on GitHub.
KAUST researchers have developed a CRISPR-Cas system using a heat-stable Cas13 protein (TccCas13a) from Thermoclostridium caenicola, compatible with RT-LAMP for rapid viral detection. The new assay, named OPTIMA-dx, enhances the specificity of RT-LAMP tests by reducing false positives in SARS-CoV-2 detection. The team, led by Dr. Magdy Mahfouz and doctoral student Ahmed Mahas, is transitioning the product to a startup phase for commercialization. Why it matters: This innovation could significantly improve point-of-care diagnostics for COVID-19 and other infections by providing a more accurate and easier-to-use testing method.
KAUST researchers are using AI to analyze satellite imagery for the automated detection of ancient stone structures in northwest Saudi Arabia, including mustatils (rectangular structures dating to the late 6th millennium BCE) and ruins in circular and triangular shapes. They developed a deep learning algorithm trained on manually identified datasets to isolate similar features over a wide area. The tool converts detected pixels into geodetic coordinates using GPS, assembling them into an online map and database. Why it matters: This project exemplifies computational archaeology, speeding up archaeological discoveries, promoting cultural heritage, and providing a methodology useful to other sectors of the economy.
This paper introduces a self-supervised learning method for point cloud analysis using an upsampling autoencoder (UAE). The model uses subsampling and an encoder-decoder architecture to reconstruct the original point cloud, learning both semantic and geometric information. Experiments show the UAE outperforms existing methods in shape classification, part segmentation, and point cloud upsampling tasks.
KAUST researchers have developed an artificial electronic retina mimicking the behavior of rod retina cells, utilizing a hybrid perovskite material (MAPbBr3) embedded in PVDF-TrFE-CEF. The photoreceptor array, made of metal-insulator-metal capacitors, detects light intensity through changes in electrical capacitance. Connected to a CMOS-sensing circuit and a spiking neural network, the 4x4 array achieved around 70 percent accuracy in recognizing handwritten numbers. Why it matters: This research paves the way for energy-efficient neuromorphic vision sensors and advanced computer vision applications, potentially revolutionizing camera technology.
This paper introduces a novel black-box adversarial attack method, Mixup-Attack, to generate universal adversarial examples for remote sensing data. The method identifies common vulnerabilities in neural networks by attacking features in the shallow layer of a surrogate model. The authors also present UAE-RS, the first dataset of black-box adversarial samples in remote sensing, to benchmark the robustness of deep learning models against adversarial attacks.
This paper introduces a self-supervised contrastive learning method for segmenting the left ventricle in echocardiography images when limited labeled data is available. The approach uses contrastive pretraining to improve the performance of UNet and DeepLabV3 segmentation networks. Experiments on the EchoNet-Dynamic dataset show the method achieves a Dice score of 0.9252, outperforming existing approaches, with code available on Github.
Researchers at KAUST have developed a nanocomposite material that converts X-rays into light with nearly 100% efficiency. The material combines a metal-organic framework (MOF) containing zirconium with an organic TADF chromophore. This design achieves high resolution and sensitivity in X-ray imaging, potentially reducing medical imaging doses by a factor of 22. Why it matters: This innovation could lead to more efficient and safer medical imaging and security screening technologies in the region and beyond.
This paper introduces a method for automatically designing convolutional neural network (CNN) architectures tailored for diabetic retinopathy (DR) diagnosis using fundus images. The approach uses k-medoid clustering, PCA, and inter/intra-class variations to optimize CNN depth and width. Validated on datasets including a local Saudi dataset and Kaggle benchmarks, the custom-designed models outperform pre-trained CNNs with fewer parameters.