A multimodal scheme for on-site maintenance resource audition and resource retrieval

The resource information and usage status in on-site maintenance of communication networks require a significant amount of identification, auditing, and retrieval work. However, there are still obstacles in resource recognition through AI image recognition, such as low recognition quality in some scenarios, difficulty in auditing and identifying the recognition results of resource data, and weak cross modal retrieval capabilities for resource data. This article proposes a multi-modal method for identifying, auditing, and retrieving on-site maintenance resources, which improves the automation and intelligence capabilities of communication network on-site maintenance.