Advanced Computer Vision Techniques for Marine Ecology

不断扩大的人类活动加上气候变化严重破坏了海洋生态系统，海洋生态系统在我们星球维持生命的能力中发挥着关键作用。然而，监测海洋健康状况的自动化技术仍然不存在，海洋科学家今天仍然需要手动处理大量原始的水下图像。这项研究旨在通过开发先进的计算机视觉工具来解决这一瓶颈，以便快速、大规模、自动地识别海洋物种。这种自动化技术预计将在空间/时间采样的速度、成本和准确性方面极大地有利于海洋生态研究，从而更好地量化海洋生态系统可容忍的环境变化程度。

Ever expanding human activity coupled with climate change has severely damaged marine ecosystems, which play a key role in our planet's ability to sustain life. Yet automated technology to monitor the health of our oceans still does not exist, with marine scientists still having today to process manually a massive amount of raw underwater imagery. This research aims to address this bottleneck by developing advanced computer vision tools for rapid, large-scale, automatic identification of marine species. Such an automated technology is expected to greatly benefit marine ecological studies in terms of speed, cost, accuracy of the spatial/temporal sampling and thus in better quantifying the level of environmental change marine ecosystems can tolerate.