Impact of Deep Learning on Localizing and Recognizing Handwritten Text in Lecture Videos

Now-a-days, the video recording technologies have turned out to be more and forceful easier utilize. Therefore, numerous universities are publishing their lectures online in order make them reachable for learners or students. These lecture videos encapsulate handwritten text written either on a paper blackboard tablet using stylus. On other hand, this mechanism of consumes huge quantity multimedia data faster manner. Thus, recognition portals has an incredibly significant demanding task. intends develop novel detection approach dataset by following four major phases, viz. (a) Text Localization, (b) Segmentation (c) Pre-processing (d) Recognition. The localization frames is initial phase here arbitrarily oriented localized Modified Region Growing (MRG) algorithm. Then, words subjected segmentation via K-means clustering, which from detected regions segmented out. Subsequently, pre-processed avoid blurriness artifacts as well. Finally, recognized Deep Convolutional Neural Network (DCNN). performance proposed model analyzed terms measures like accuracy, precision, sensitivity specificity exhibit supremacy video. Experimental results reveal that at Learning Percentage 70, presented work highest accuracy 89.3% 500 count frames.