To scientifically assess the status and spatial distribution of fish resources in the Gutang waters of Poyang Lake， supporting the conservation and resource recovery of endangered species such as Coilia nasus.In July 2024， a dual-frequency identification sonar （DIDSON） was used for acoustic surveys in the study area. Fish targets were identified and extracted by combining an Echoview-based recognition model and a YOLO deep learning detection model. Catch statistics were integrated to estimate fish density， abundance， biomass， and spatial distribution patterns.A total of 1 384 fish targets were detected. The Echoview model achieved an average accuracy of 91.91%， while the YOLO model attained average precision and recall rates of 96.4% and 88.5%， respectively. The total fish abundance was estimated at approximately 135 200 individuals， with a total biomass of about 165 000 kg. Spatially， fish were predominantly distributed in nearshore areas horizontally and concentrated at depths of 5-10 m vertically.The integrated approach combining DIDSON， Echoview， and YOLO proved effective for high-accuracy fish target recognition and resource estimation， with the YOLO model offering greater objectivity and processing efficiency This study provides a technical reference for intelligent fish target recognition in sonar images and offers a scientific basis for the management and conservation of fish resources in the Gutang waters of Poyang Lake.