With the evolution of mobile devices, the volume of data that each user has attached to him is increasingly noticeable. To ensure the integrity of data and storage capacity, cloud platforms enable users to store and synchronize data between their devices, resulting in users commonly having thousands of images on this type of platforms. Therefore, this has brought the challenge of carrying out efficient searches for personal images. We propose a system which is easily integrated into cloud platforms and provides fast image searches for the user. The solution focuses on semantic search by similarity of textual description based on the use of deep learning models that have the ability to understand semantic relationships between text and images. To achieve more accurate results, traditional approaches are adopted were results are filtered using textual annotations generated by a model capable of performing multi-modal tasks. Filtering is carried out using another model, which analyzes the semantics of textual data. The proposed system aims to produce accurate and filtered results according to the user’s needs. With this approach, it is possible to conduct a precise search that goes beyond searching by date or file type.