Browsing by Author "Mule, H."

Filter results by typing the first few letters
Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    Handwritten Text Recognition from an Image with Android Application
    (Institute of Electrical and Electronics Engineers Inc., 2022) Mule, H.; Kadam, N.; Naik, D.
    Nowadays, Storing information from handwritten documents for future use is becoming necessary. An easy way to store information is to capture handwritten documents and save them in image format. Recognizing the text or characters present in the image is called Optical Character Recognition. Text extraction from the image in the recent research is challenging due to stroke variation, inconsistent writing style, Cursive handwriting, etc. We have proposed CNN and BiLSTM models for text recognition in this work. This model is evaluated on the IAM dataset and achieved 92% character recognition accuracy. This model is deployed to the Firebase as a custom model to increase usability. We have developed an android application that will allow the user to capture or browse the image and extract the text from the picture by calling the firebase model and saving text in the file. To store the text file user can browse for the appropriate location. The proposed model works on both printed and handwritten text. © 2022 IEEE.
  • No Thumbnail Available
    Item
    Video to Text Generation Using Sentence Vector and Skip Connections
    (Springer, 2023) Mule, H.; Naik, D.
    Nowadays, video data is increasing rapidly and the need of robust algorithms to process the interpretation of the video. A textual alternative will be more effective and save time. We aim to produce the caption for the video. The most famous architecture used for this is the encoder-decoder (E-D) model. Recent attempts have focused on improving performance by including 3D-CNN, transformers, or structural changes in the basic LSTM units used in E-D. Sentence vectors are used in this work, improving the E-D model’s performance. From the video file, a sentence vector is generated and used by the decoder to generate an accurate description by using previously generated words. Skip connection in the encoder part avoids the vanishing gradients problem. All of our studies use the MSVD and CHARADES datasets. Four famous metrics, BLEU@4, METEOR, ROUGE, and CIDER, are used for performance evaluation. We have compared the performance of BERT, ELMo, and GloVe word embeddings. On experimental analysis, BERT embedding outperformed the ELMo and GloVe embeddings. For feature extraction, pretrained CNNs, NASNet-Large, VGG-16, Inception-v4, and Resnet152 are used, and NASNet-Large outperformed other models. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.