Caltech Fish Counting Dataset (CFC)
收藏资源简介:
Caltech Fish Counting Dataset(CFC)是由加州理工学院创建的大型数据集,专注于通过声纳视频检测、跟踪和计数鱼类。该数据集包含超过1500个视频,总计超过半百万的标注,主要用于推动低信噪比环境下的计算机视觉应用和多目标跟踪(MOT)及计数领域的泛化研究。CFC数据集的独特之处在于其数据来源于自然环境,其中目标不易解析,外观特征难以用于目标重新识别,为研究者提供了一个挑战性的基准,以评估和改进算法在未知测试地点的泛化性能。此外,该数据集对于保护生态学等领域具有重要影响,特别是在可持续渔业管理方面。
Caltech Fish Counting Dataset (CFC) is a large-scale dataset created by the California Institute of Technology, focusing on detecting, tracking and counting fish via sonar videos. This dataset contains over 1,500 videos with a total of more than half a million annotations, and is primarily used to advance computer vision applications in low signal-to-noise ratio (SNR) environments as well as generalizable research in the fields of multi-object tracking (MOT) and counting. A unique feature of the CFC dataset is that its data is sourced from natural environments, where fish targets are difficult to discern and their appearance features are not conducive to target re-identification. It provides researchers with a challenging benchmark for evaluating and improving the generalization performance of algorithms in unseen test sites. Furthermore, this dataset has important implications for fields such as conservation ecology, especially in sustainable fisheries management.




