The second dAIEDGE Open Call concluded with outstanding results, exceeding all our expectations.
After two months of extensive outreach and collaboration with the community, including two informative webinars attended by 138 participants, we received an extraordinary response, reflecting the growing interest in artificial intelligence applied to Edge Computing.
dAIEDGE 2OC Challenges
Challenges from the dAIEDGE consortium that drive research and development activities aimed at solving key industrial problems using advanced AI and edge AI technologies.
AETHON ENGINEERNG SINGLE MEMBER P.C.
Challenge: 19
Acronym: MERCAI
A blockchain-powered marketplace enabling AI developers to publish, train, and monetize models while earning incentives through decentralized smart contracts.
CONSORTIUM: Universidade de Vigo (University of Vigo) & Logicmelt Technologies SL
Challenge: 10
Acronym: QUAD
Quantized Implementation of Deep Convolution Neural Networks for Vehicle Detection on Field Programmable System-on-Chip (FPSoC) devices using high level Python-based PYNQ framework
GradeBuilder SL
Challenge: 9
Acronym: Simfero
A hybrid AI memory system for off-chip weight optimization, enabling efficient knowledge retention in educational LLMs with reduced computational overhead
INLECOM INNOVATION ASTIKI MI KERDOSKOPIKI ETAIREIA
Challenge: 16
Acronym: AMFITRITE
AMFITRITE revolutionizes HAB detection with a robust semi-automated annotation method and CS6-compatible CNNs, enhancing accuracy and enabling scalable, global low-latency monitoring of HABs.
PLAIXUS P.C.
Challenge: 12
Acronym: NAVIR
Low-power, real-time multimodal SNNs for robust human-robot interaction in noisy settings
PLEGMA LABS TECHNOLOGIKES LYSEIS ANONYMOS ETAIRIA
Challenge: 1
Acronym: CANDLE
Privacy-preserving concealed weapon detection with lightweight optimized AI models and thermal frame fusion with face blurring on mobile edge devices
Radio Analog Micro Electronics srl
Challenge: 11
Acronym: ViTFireEdge Accelerator
Pioneering low-power, real-time Vision Transformer acceleration on RISC-V SoC FPGAs for advanced edge-based image classification.
Software Competence Center Hagenberg GmbH
Challenge: 13
Acronym: E*3
Efficient-Edge-Embeddings (E*3) develops a multi-objective optimization approach for running text embedding models directly on edge devices, balancing energy use, latency, and accuracy. The project validates energy-efficient NLP inference across multiple sentence transformers and edge hardware platforms, targeting significant power savings without degrading performance
Stroma Teknoloji Sanayi ve Ticaret Limited Sirketi
Challenge: 3
Acronym: Krait
A JAX-first performance optimization and edge-deployment pipeline that cuts memory and latency in the active-inference inference/control hot paths, then exports the optimized model to ONNX and builds TensorRT engines for real-time execution on NVIDIA Jetson
University College Dublin, National University of Ireland
Challenge: 15
Acronym: TITAN
TITAN: Privacy-Preserving Distributed LLM Training for Edge Devices, Enabling Secure and Cost-Effective AI Solutions Across Industries.
