基于视觉的渔船监管关键技术研究与系统开发

Key Technology and System Development of Vision-based Fishing Vessel Supervision

当前小型渔船监管的信息化水平不高,依赖人工巡查监管存在漏检误检,易导致渔业生产安全隐患。针对此问题,首先,构建了小型渔船的进出港目标识别以及人员安全检查的系列深度学习模型;其次,开发了一套渔船进出港监管系统,实现了渔业生产管理的智能化;最后,将监管系统成功部署,促进了烟台长岛渔业生产监管的智能化。该系统实现了渔船及人员目标检测与轨迹追踪、船牌识别、救生衣穿着检测等功能,并与年审渔船信息以及违禁出海管理等对接,实现对渔船未年审出海、违章载客、未穿救生衣、违禁出海等违规活动的告警。研究工作提高了渔船进出港监管效率,降低了渔业生产的安全隐患。

The intelligence in the supervision of small fishing boats is limited at present,relying on manual inspection methods,which may result in omissions and errors,posing potential safety hazards in fisheries.In this paper,firstly,several deep learning models are constructed for the supervision of small fishing vessels.Secondly,an entry and exit supervision system for fishing vessels is developed,achieving the intelligence of fisheries management.Finally,the supervision system was successfully deployed,promoting the intelligence of fisheries production supervision in Changdao,Yantai.The system integrates the models of target detection of fishing vessels and personnel,identification of vessel license plates,life jacket wearing detection,and interfaces with the information of annual examination of fishing vessels and the database of prohibited periods of fishing vessels.