Forest resources are the most precious natural resources of human beings. According to the national strategic planning and guidance, China's forest area increases continuously, quality rises steadily, and efficiency increases continuously. In recent years, the deep integration of artificial intelligence technology represented by deep learning and forestry production management and processing industry is one of the important trends to realize the green and intelligent development of forestry. This paper summarizes the research progress of target detection, recognition and classification based on different deep learning algorithms and models in forestry production management and processing, comprehensively compares the advantages and disadvantages of different model algorithms, and puts forward some research suggestions, such as establishing multi-source database, optimizing algorithm model, improving hardware configuration, etc., which provides a reference for the intelligent development of forestry industry.