Automatic text detection and recognition in images and videos have emerged and aroused widespread interest in recent years due to the dramatic growth of visual information. It seems that there is the lack of any effective model for the Farsi text detection in images. In this paper, a new framework is proposed for the Farsi text detection and localization using the up-to-date real-time object detection framework YOLOv5 in videos and images. To evaluate the novel model, a new dataset of news videos is collected. Experimental results show that the proposed model achieves quite promising performance on the new dataset.
