 This paper proposes a novel end-to-end system for detecting and generating text from Bengali signed characters. The system consists of two phases. Firstly, a YOLO V4 tiny model is used to detect 49 different signed characters, including 36 Bengali alphabet characters, 10 numeric characters, and 3 special characters. Then, a long short term memory, LSTM, model is used to generate meaningful text from the detected characters. The proposed system achieves a map of 99.7% on the BDSL 49 dataset, which contains approximately 14,745 images of 49 different classes. Additionally, the authors compared the performance of three different YOLO V4 models and found that the YOLO V4 tiny model outperformed both YOLO V4 and YOLO V7 models. This article was authored by Nassima Begum, Rashik Rahman, Nusrat Jahan, and others. We are article.tv. Links in the description below.