A Hybridized Deep Learning Method for Bengali Image Captioning

Researchgate: researchgate.net/A Hybridized Deep Learning Method for Bengali Image Captioning

An omnipresent challenging research topic in computer vision is the generation of captions from an input image. Previously, numerous experiments have been conducted on image captioning in English but the generation of the caption from the image in Bengali is still sparse and in need of more refining. Only a few papers untill now have worked on image captioning in Bengali. Hence, we proffer a standard strategy for Bengali image caption generation on two different sizes of the Flickr8k dataset and BanglaLekha dataset which is the only publicly available Bengali dataset for image captioning. Afterward, the Bengali captions of our model were compared with Bengali captions generated by other researchers using different architectures. Additionally, we employed a hybrid approach based on InceptionResnetV2 or Xception as Convolution Neural Network and Bidirectional Long Short-Term Memory or Bidirectional Gated Recurrent Unit on two Bengali datasets. Furthermore, a different combination of word embedding was also adapted. Lastly, the performance was evaluated using Bilingual Evaluation Understudy and proved that the proposed model indeed performed better for the Bengali dataset consisting of 4000 images and the BanglaLekha dataset.

Bengali Image Captioning with Visual Attention

Researchgate: researchgate.net/Bengali Image Captioning with Visual Attention

Attention based approaches has been manifested to be an effective method in image captioning. However, attention can be used on text called semantic attention or on image which in known as spatial attention. We chose to implement the latter as the main problem of image captioning is not being able to detect objects in image properly. In this work, we develop an approach which extracts features from images using two different convolutional neural network and combines the features with an attention model in order to generate caption with an RNN. We adapted Xception and InceptionV3 as our CNN and GRU as our RNN. Moreover, we Evaluated our proposed model on Flickr8k dataset translated into Bengali. So that captions can be generated in Bengali using visual attention.

Bornon: Bengali Image Captioning with Transformer-based Deep learning approach

Researchgate: researchgate.net/Bengali Image Captioning with Visual Attention

Image captioning using Encoder-Decoder based approach where CNN is used as the Encoder and sequence generator like RNN as Decoder has proven to be very effective. However, this method has a drawback that is sequence data must be processed in order. The beginning of the sequence must be processed before the end. To solve this issue we utilized the transformer-based model to caption images in Bengali. Unlike other sequence models, the Transformer uses an attention mechanism that provides context for any position in the input sequence. For instance, if the input data is a natural language sentence, the transformer does not need to process the beginning of the sentence before the end. As a result, we utilized three different Bengali datasets to generate Bengali captions from images using the Transformer model. Additionally, we fine-tuned the number of heads and layers in the transformer of all three Bengali datasets. Finally, we compared the result of the transformer-based model with a visual attention-based approach to caption images in Bengali.

Automated Bengali abusive text classification: Using deep learning techniques

In recent years, the breakdown of most human causes that hampers a large number of people is verbal abuse. The most noticeable fact is that most of the verbal abuse is made online by people behind the screen. While this problem has occurred with growing modern technologies, a modern solution is needed to solve the drawback. The growing rate of abusive comments and hate speech online is rapidly increasing on a large scale, and manual reports and corrections cannot help this critical situation. This proposed model introduces automation of the hate speech filtering process through deep learning. This work uses only Bengali datasets to find the real classification, as many Bengali and English mixed research works are available in this field. Using the intelligent automated classification of text comments in a limited resource-constrained language (example: Bengali) is critical for several reasons. This proposed system classifies 'Personal Offensive Text', 'Geographical Offensive Text', 'Religious Offensive text', 'Crime Offensive Text', 'Entertainment, Sports, Meme Tiktok, and others'. By combining two large datasets for this research and employing Bengali BERT, the best feasible outcome has been achieved. This paper indicates the complete results of the combined two different datasets and classified into nine segments, while the proposed Bengali BERT models achieved the highest accuracy of 0.706 and a weighted f1 score of 0.705 in the identification and classification tasks.