With the development of fishery industry,accurate estimation of the number of fish in aquaculture waters is of great importance to fish behavior analysis,bait feeding and fishery resource investigation.In this paper,w...With the development of fishery industry,accurate estimation of the number of fish in aquaculture waters is of great importance to fish behavior analysis,bait feeding and fishery resource investigation.In this paper,we propose a method for fish density estimation based on the multi-scale context enhanced convolutional network,which could map a fish school image taken at any angle to a density map,and calculate the number of fish in the image finally.In order to eliminate the influence of camera perspective effect and image resolution on density estimation,multi-scale filters are utilized in a convolutional neural network to process fish image in parallel.And then,the context enhancement module is merged in the network structure to help the network understand the global context information of the image.Finally,different feature maps are merged together to construct the density map of fish school images,and finally get the number of fish in the image.In order to make the effectiveness of our method valid,we test the proposed method on DlouDataset.The results show that the proposed method has lower mean square error and mean absolute error,which is helpful to improve the accuracy of the fish counting in dense fish school images.展开更多
We present a lightweight and efficient semisupervised video object segmentation network based on the space-time memory framework.To some extent,our method solves the two difficulties encountered in traditional video o...We present a lightweight and efficient semisupervised video object segmentation network based on the space-time memory framework.To some extent,our method solves the two difficulties encountered in traditional video object segmentation:one is that the single frame calculation time is too long,and the other is that the current frame’s segmentation should use more information from past frames.The algorithm uses a global context(GC)module to achieve highperformance,real-time segmentation.The GC module can effectively integrate multi-frame image information without increased memory and can process each frame in real time.Moreover,the prediction mask of the previous frame is helpful for the segmentation of the current frame,so we input it into a spatial constraint module(SCM),which constrains the areas of segments in the current frame.The SCM effectively alleviates mismatching of similar targets yet consumes few additional resources.We added a refinement module to the decoder to improve boundary segmentation.Our model achieves state-of-the-art results on various datasets,scoring 80.1%on YouTube-VOS 2018 and a J&F score of 78.0%on DAVIS 2017,while taking 0.05 s per frame on the DAVIS 2016 validation dataset.展开更多
基金This work is supported by Institute of Marine Industry Technology of Universities in Liaoning Province(2018-CY-34)National Natural Science Foundation of China(31972846)+1 种基金China Postdoctoral Science Foundation(2018M640239)Acknowledgement for the Data Support from National Marine Science Data Center(Dalian),National Science&Technology Resource Sharing Service Platform of China(http://odc.dlou.edu.cn/).
文摘With the development of fishery industry,accurate estimation of the number of fish in aquaculture waters is of great importance to fish behavior analysis,bait feeding and fishery resource investigation.In this paper,we propose a method for fish density estimation based on the multi-scale context enhanced convolutional network,which could map a fish school image taken at any angle to a density map,and calculate the number of fish in the image finally.In order to eliminate the influence of camera perspective effect and image resolution on density estimation,multi-scale filters are utilized in a convolutional neural network to process fish image in parallel.And then,the context enhancement module is merged in the network structure to help the network understand the global context information of the image.Finally,different feature maps are merged together to construct the density map of fish school images,and finally get the number of fish in the image.In order to make the effectiveness of our method valid,we test the proposed method on DlouDataset.The results show that the proposed method has lower mean square error and mean absolute error,which is helpful to improve the accuracy of the fish counting in dense fish school images.
基金partially supported by the National Natural Science Foundation of China(Grant Nos.61802197,62072449,and 61632003)the Science and Technology Development Fund,Macao SAR(Grant Nos.0018/2019/AKP and SKL-IOTSC(UM)-2021-2023)+1 种基金the Guangdong Science and Technology Department(Grant No.2020B1515130001)University of Macao(Grant Nos.MYRG2020-00253-FST and MYRG2022-00059-FST).
文摘We present a lightweight and efficient semisupervised video object segmentation network based on the space-time memory framework.To some extent,our method solves the two difficulties encountered in traditional video object segmentation:one is that the single frame calculation time is too long,and the other is that the current frame’s segmentation should use more information from past frames.The algorithm uses a global context(GC)module to achieve highperformance,real-time segmentation.The GC module can effectively integrate multi-frame image information without increased memory and can process each frame in real time.Moreover,the prediction mask of the previous frame is helpful for the segmentation of the current frame,so we input it into a spatial constraint module(SCM),which constrains the areas of segments in the current frame.The SCM effectively alleviates mismatching of similar targets yet consumes few additional resources.We added a refinement module to the decoder to improve boundary segmentation.Our model achieves state-of-the-art results on various datasets,scoring 80.1%on YouTube-VOS 2018 and a J&F score of 78.0%on DAVIS 2017,while taking 0.05 s per frame on the DAVIS 2016 validation dataset.
