The problem of foreign objects attached to transmission lines is becoming increasingly prominent, threatening the safe and stable operation of power grid. The traditional manual inspection method is not suitable for the maintenance of modern power grid due to its low efficiency and high missing rate, while the rapid development of deep learning technology provides new possibilities for the field of target detection. This paper provides a survey of the most current advancements in foreign object detection techniques for transmission lines using deep learning. This paper presents a thorough examination and assessment of current datasets. It offers readers a detailed summary of available data sources. In addition, this paper discusses existing detection techniques and their constraints. This allows readers to acquire a comprehensive understanding of the most recent technological advancements in this area. Meanwhile the paper identifies future research areas that may lead to significant technological advancements for industry researchers.