A5025961783- Advanced Image and Video Retrieval Techniques
- Generative Adversarial Networks and Image Synthesis
- Education and Learning Interventions
- Sleep and Work-Related Fatigue
- Gaze Tracking and Assistive Technology
- Image and Video Quality Assessment
- Advanced Neural Network Applications
- Computability, Logic, AI Algorithms
- Advanced Data Processing Techniques
- Advanced Image Processing Techniques
- Face recognition and analysis
- Emotion and Mood Recognition
- Educational Research and Pedagogy
- Educational Technology and Assessment
- Image and Signal Denoising Methods
- Innovative Teaching and Learning Methods
- Big Data and Business Intelligence
- Technology and Human Factors in Education and Health
- Image Processing and 3D Reconstruction
- EEG and Brain-Computer Interfaces
Kanazawa University
2022-2024
Xiamen University of Technology
2023
In recent years, the rapid prevalence of high-definition video in Internet Things (IoT) systems has been directly facilitated by advances imaging sensor technology. To adapt to limited uplink bandwidth, most media platforms opt compress videos bitrate streams for transmission. However, this compression often leads significant texture loss and artifacts, which severely degrade Quality Experience (QoE). We propose a latent feature diffusion model (LFDM) compressed quality enhancement,...
The complexity of deep neural network models (DNNs) severely limits their application on devices with limited computing and storage resources. Knowledge distillation (KD) is an attractive model compression technology that can effectively alleviate this problem. Multi-teacher knowledge (MKD) aims to leverage the valuable diverse distilled by multiple teacher networks improve performance student network. Existing approaches typically rely simple methods such as averaging prediction logits or...
Due to the high flexibility and conformity people’s usage habits, text description has been widely used in image synthesis research recently achieved many encouraging results. However, can only determine basic content of generated cannot specific shape synthesized object, which leads poor practicability. More importantly, current text-to-image use new descriptions further modify result. To solve these problems, this paper proposes a text-guided customizable manipulation method. The proposed...