A5003581478- Advanced Chemical Sensor Technologies
- Solid State Laser Technologies
- Advanced Sensor and Control Systems
- Hand Gesture Recognition Systems
- Handwritten Text Recognition Techniques
- Laser-Matter Interactions and Applications
- Neural Networks and Applications
- Topic Modeling
- Astrophysics and Cosmic Phenomena
- Natural Language Processing Techniques
- Gamma-ray bursts and supernovae
- Neutrino Physics Research
- Vehicle License Plate Recognition
- Laser-Plasma Interactions and Diagnostics
Harbin Institute of Technology
2023
Traffic sign is a critical map feature for navigation and traffic control. Nevertheless, current methods recognition rely on traditional deep learning models, which typically suffer from significant performance degradation considering the variations in data distribution across different regions. In this paper, we propose TSCLIP, robust fine-tuning approach with contrastive language-image pre-training (CLIP) model worldwide cross-regional recognition. We first curate benchmark dataset by...
The Gravitational Wave Burst High-energy Electromagnetic Counterpart All-sky Monitor (GECAM), consists of 2 small satellites that each contain 25 LaBr3 (lanthanum bromide doped with cerium chloride) detectors and 8 plastic scintillator detectors. detector signals are read out using a silicon photomultiplier (SiPM) array. In this study, an acquisition algorithm for in-orbit real-time SiPM array data is designed implemented, the output event packet defined. Finally, algorithm's efficacy verified.
The number of parameters in large-scale language models based on transformers is gradually increasing, and the scale computing clusters also growing. technology quickly mobilizing large amounts resources for parallel becoming increasingly important. In this paper, we propose an automatic algorithm that automatically plans strategy with maximum throughput model hardware information. By decoupling training time into computation, communication, overlap, established a duration simulation model....
Based on the experimental parameters in 2.0 PW Ti:Sapphire laser system, we propose a method to suppress parasitic lasing effects large-aperture amplifiers by using temporal dual-pulse pump.
High-precision vehicle velocity measurement methods typically require additional context, such as the rough location of target vehicle. In Internet Vehicles domain, it is possible to achieve real-time based on characteristics wireless communication signals without any extra context. this paper, we first propose a algorithm utilizing convolutional neural network (CNN) with differential constellation diagram signal input. We also develop an acceleration gated recurrent units (GRUs) that...