On February 24, the paper of Professor Wang Peng's research group from the School of Computer Science of Northwestern Polytechnical University was adopted by the 2020 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). CVPR was first held in 1983 by IEEE, and is a top-level conference in the field of computer vision and pattern recognition. This year, CVPR received a total of more than 7,000 submissions, with 6,656 valid submissions. Among which, 1,470 submissions were adopted, representing an adoption rate of 22.0%.

The paper entitled "NAS-FCOS: Fast Neural Architecture Search for Object Detection" was jointly completed by NPU and The University of Adelaide in Australia. First-year graduate students Wang Ning and Gao Yang from NPU are the co-lead authors of this paper. This paper mainly introduces the work of the these two graduates in Neural Architecture Search during their undergraduate graduation design. Neural Architecture Search is one of the hot research directions in the field of deep learning in recent years. Its goal is to use machine learning instead of artificial design of neural network structure to help researchers get rid of tedious parameter adjustment engineering. This paper mainly introduces a fast neural architecture search framework based on reinforcement learning. Through carefully designed search space and strategy, a good network structure can be obtained in target detection tasks, with both accuracy and operation efficiency guaranteed. The main body of the thesis was completed by Wang Ning and Gao Yang during the undergraduate graduation design period, and submitted after enriching and further revising the graduation design in the first semester of the research.

Wang Peng works as a professor and doctoral supervisor in the School of Computer Science of NPU, and belongs to the National Space-air-land-sea Big Data Integration Application Technology Engineering Lab. His undergraduate or junior graduate students have published many articles in well-known journals and achieved the outstanding results in various artificial intelligence competitions at home and abroad. The specific research directions include vision-language interaction, text image recognition, neural architecture search, target re-recognition, target detection and tracking, etc. He has published more than 20 papers in CCF A journals and conferences such as TPAMI, IJCV, TIP, CVPR, ICCV, AAAI, IJCAI, etc.

(Written by: Wang Ning, Gao Yang, reviewed by: Gao Wu)