URPC Competition: Underwater Target Detection
Competition Profile: In recent years, with the development of deep learning theory and algorithm, the accuracy of target detection algorithm has also been significantly improved. However, due to the complex underwater imaging environment and illumination conditions, the underwater image often has the problems of low contrast, texture distortion and uneven illumination. The development of underwater target detection is slow, and there is still a lack of large-scale underwater training data set in real environment. The competition aims at attracting researchers in the field of multimedia/computer vision in China. It sincerely hopes that researchers from academia and industry will take part in it. It aims to promote the development of artificial intelligence and the technological progress of underwater vehicles, promote the development of underwater target perception and detection theory, technology and application, and enhance related research level.
1. Training: Provide underwater image training set, but not limited training set;
2. Testing: Providing underwater image test set;
3. Preliminary competition: All the image data in this competition are real underwater image data. The result of the competition is based on the accuracy. Participants are required to submit the results. The organizer tests the code and gets the final result. All the evaluation results are published online on time.
4. Quarter-final: The Organizer provides the machine, the team configures the code program, reproduces the preliminary results, and compares the algorithm speed.
Competition data (training data):
Link: https://pan.baidu.com/s/1Ec-1-YiKpGaOcYfirkD2PQ
Extraction code: ugr9
Schedule:
Test data set release time: 2019.8.8
Final result submission time: 2019.8.15-2018.8.16
Competition Technology Group:
Group leader: Huang Qingming and Song Su
Vice Team Leaders: Cui Junhong and Lu Huchuan
Members: Wang Hongyu, Shen Hengtao, Baixiang, Li Haojie, Yang Zheng, Jiang Yugang, Geng Xin, Tang Jinhui, Huanghua and Yu Junzhi
Contacts:
Wang Dong, wdice@dlut.edu.cn
