This paper proposes a model for analyzing factors affecting goodwill impairment based on computer vision and uniform manifold approximation and projection (UMAP). The model combines computer vision technology and mines the key factors affecting goodwill value by processing multi-dimensional information of enterprise-related images, texts and financial data. The UMAP dimensionality reduction technology is used to ensure that the structure of the data in the low-dimensional space remains uniformly distributed, and feature extraction and classification prediction are performed through a multi-layer neural network. In addition, this paper analyzes the feature changes of different enterprises before and after goodwill impairment through model simulation, and verifies the accuracy and stability of the algorithm. The simulation results show that the model can effectively identify the key factors affecting goodwill impairment with an accuracy of more than 95%, and has good applicability to different types of enterprises. This paper provides a new solution to the problem of goodwill impairment and a scientific basis for enterprise managers to make decisions.