TY  - JOUR
AU  - Khan, Abdullah
AU  - Singh, Jaswinder
PY  - 2025
DA  - 2025/08/01
TI  - A Novel Image Captioning Technique Using Deep Learning Methodology
JO  - ICCK Transactions on Machine Intelligence
T2  - ICCK Transactions on Machine Intelligence
JF  - ICCK Transactions on Machine Intelligence
VL  - 1
IS  - 2
SP  - 52
EP  - 68
DO  - 10.62762/TMI.2025.886122
UR  - https://www.icck.org/article/abs/TMI.2025.886122
KW  - convolutional neural networks (CNN)
KW  - recurrent neural networks (RNN)
KW  - deep learning
KW  - image captioning
KW  - LSTM
KW  - DenseNet201
KW  - attention mechanism
KW  - BLEU scor
KW  - natural language processing (NLP)
KW  - multimodal learning
KW  - content retrieval
AB  - The capacity of AI systems to generate captions for images autonomously represents a significant advancement in artificial intelligence and language understanding. This paper presents an advanced image captioning system that employs deep learning techniques, particularly convolutional neural networks (CNNs) and recurrent neural networks (RNNs), to produce contextually appropriate and meaningful descriptions of visual content. The proposed method extracts features using the DenseNet201 model, enabling a more comprehensive and hierarchical understanding of image components. These extracted features are then fed into a long short-term memory (LSTM) network, a specialized RNN variant designed to capture sequential dependencies in language, yielding coherent and fluent captions. The model is trained and evaluated on the well-known Flickr8k dataset, achieving competitive performance as measured by BLEU score metrics and demonstrating its ability to generate human-like descriptions. This integration of CNNs and RNNs highlights the effectiveness of combining computer vision and natural language processing for automated caption generation. The approach has potential applications across various domains, including assistive technologies for the visually impaired, automated content creation for digital media, enhanced indexing and retrieval of multimedia assets, and improved human-computer interaction. Furthermore, advancements in attention mechanisms and transformer-based models present opportunities to further enhance the accuracy and contextual relevance of image captioning systems. The study underscores the broader implications of machine-generated captions for improving accessibility, boosting searchability in large-scale databases, and enabling seamless AI-human collaboration in content interpretation and storytelling.
SN  - 3068-7403
PB  - Institute of Central Computation and Knowledge
LA  - English
ER  - 

@article{Khan2025A,
  author = {Abdullah Khan and Jaswinder Singh},
  title = {A Novel Image Captioning Technique Using Deep Learning Methodology},
  journal = {ICCK Transactions on Machine Intelligence},
  year = {2025},
  volume = {1},
  number = {2},
  pages = {52-68},
  doi = {10.62762/TMI.2025.886122},
  url = {https://www.icck.org/article/abs/TMI.2025.886122},
  abstract = {The capacity of AI systems to generate captions for images autonomously represents a significant advancement in artificial intelligence and language understanding. This paper presents an advanced image captioning system that employs deep learning techniques, particularly convolutional neural networks (CNNs) and recurrent neural networks (RNNs), to produce contextually appropriate and meaningful descriptions of visual content. The proposed method extracts features using the DenseNet201 model, enabling a more comprehensive and hierarchical understanding of image components. These extracted features are then fed into a long short-term memory (LSTM) network, a specialized RNN variant designed to capture sequential dependencies in language, yielding coherent and fluent captions. The model is trained and evaluated on the well-known Flickr8k dataset, achieving competitive performance as measured by BLEU score metrics and demonstrating its ability to generate human-like descriptions. This integration of CNNs and RNNs highlights the effectiveness of combining computer vision and natural language processing for automated caption generation. The approach has potential applications across various domains, including assistive technologies for the visually impaired, automated content creation for digital media, enhanced indexing and retrieval of multimedia assets, and improved human-computer interaction. Furthermore, advancements in attention mechanisms and transformer-based models present opportunities to further enhance the accuracy and contextual relevance of image captioning systems. The study underscores the broader implications of machine-generated captions for improving accessibility, boosting searchability in large-scale databases, and enabling seamless AI-human collaboration in content interpretation and storytelling.},
  keywords = {convolutional neural networks (CNN), recurrent neural networks (RNN), deep learning, image captioning, LSTM, DenseNet201, attention mechanism, BLEU scor, natural language processing (NLP), multimodal learning, content retrieval},
  issn = {3068-7403},
  publisher = {Institute of Central Computation and Knowledge}
}