Research on Key Technologies of Power Equipment Target Recognition Based on Visual Large Model

Based on high-precision pre-training segmentation large model training technology, this paper uses model fine-tuning technology, pre-training segmentation large model training technology and channel pruning segmentation large model compression technology based on the independence of feature maps to carry out high-efficiency model training, forming a key component identification model of transmission lines and power grid towers, effectively improving the automation and intelligence level of power grid equipment operation and maintenance. This method reduces the input cost of intelligent inspection.