Prof. Ernest Valveny
Dept. CiŤncies Computaciů - Computer Vision Center
Universitat AutÚnoma de Barcelona
Read and reason: the role of text in semantic image interpretation
Textual content in human environments conveys important high-level semantic information that is not available in any other form in the scene. Interpreting written information in human environments is necessary for performing most everyday tasks like making a purchase, using public transportation, finding a place in the city, etc. Therefore, being able to recognize textual information in images is essential for many high-level reasoning tasks on images. With recent advances in machine learning using deep neural networks, text detection and recognition methods have reached appealing results in complex real life scenarios. In this talk we will review these recent developments in text detection and recognition and we will explore how recognized scene text can be leveraged and combined with visual information in order to address several image semantic interpretation tasks, such as classification, retrieval, captioning or question answering.
Lloc: Sala de Graus ETSE