KAUST Professor Wolfgang Heidrich is researching computational imaging systems that jointly design optics and image reconstruction algorithms. He focuses on hardware-software co-design for imaging systems with applications in HDR, compact cameras, and hyperspectral imaging. Heidrich's work on HDR displays was the basis for Brightside Technologies, acquired by Dolby in 2007. Why it matters: This research aims to advance imaging technology through AI-driven design, potentially impacting various fields from consumer electronics to scientific research within the region and globally.
The paper introduces the Prism Hypothesis, which posits a correspondence between an encoder's feature spectrum and its functional role, with semantic encoders capturing low-frequency components and pixel encoders retaining high-frequency information. Based on this, the authors propose Unified Autoencoding (UAE), a model that harmonizes semantic structure and pixel details using a frequency-band modulator. Experiments on ImageNet and MS-COCO demonstrate that UAE effectively unifies semantic abstraction and pixel-level fidelity, achieving state-of-the-art performance.
KAUST and Saudi healthtech company amplifAI health have signed an MoU to develop a new disease detection system. The system will combine amplifAI's AI technology with KAUST's HyplexTM hyperspectral imaging, initially for diabetic foot complications. Clinical trials are planned, with aims to reduce amputations and save Saudi Arabia over 2 billion Riyals annually. Why it matters: This partnership showcases the potential of combining Saudi AI and advanced imaging technologies to address pressing healthcare challenges in the region, particularly diabetes.
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.
MBZUAI researchers have developed GeoPixel, a new multimodal model for pixel grounding in remote sensing images. GeoPixel associates individual pixels with object categories, enabling detailed image analysis by linking language to objects at the pixel level. The model was trained on a new dataset and benchmark, outperforming existing systems in precision. Why it matters: This advancement enhances the utility of remote sensing data for critical applications like environmental management and disaster response by providing more granular and accurate image interpretation.
KAUST's Hydrology and Land Observation (Halo) lab, led by Matthew McCabe, is using drones and satellites to monitor agricultural water usage in Saudi Arabia. They employ thermal cameras, sensors, and imagery from CubeSats to map crop types, health, and water stress. The team uses machine learning and AI to analyze the images, aiming to promote sustainable water management. Why it matters: This research addresses critical water scarcity issues in the region by providing data-driven insights for more efficient agricultural practices.
KAUST's Hydrology, Agriculture and Land Observation Lab (HALO) uses remote sensing tools like drones to map and monitor agricultural sites, including one of the world's largest olive farms in Al Jouf. They also collaborate with the Ministry of Environment, Water and Agriculture of Saudi Arabia and the Center for Desert Agriculture. HALO's mangrove project focuses on carbon sequestration by Avicennia marina mangroves in the Red Sea as part of the KAUST Circular Carbon Initiative. Why it matters: This research contributes to understanding and promoting the conservation of mangrove ecosystems for carbon neutrality in line with Saudi Arabia's sustainability goals.