LATEST PROJECTS
Natural Language Descriptions of Videos: An Analytical Study
Ayush Sharma, Udit Saxena, Shubham Mukherjee
Real world videos around us are composed of complex frames which have either simple or complex temporal relationships between the frames. The length and quality of such videos are often variable. So, it poses a challenging task for researchers and analysts around the world to represent the information about the content in a video with a small text of words. The other half of this challenge is to be selective about the information captured and only consider the salient features of it. In this project, we have looked at the problem of video captioning and have presented the results of our experiments.
Project report link.
Language script identification using bag of visual words
Ayush Sharma, Abhyudai Nouni
Script and language identification are precursors to modern multilingual optical character recognition (OCR). The task of visual script recognition is often non-trivial since we are analyzing images containing text and not the text itself. It poses an interesting task for researchers to find ways to represent the uniqueness of a script in tangi- ble terms and to compute differences or similarity between these representations. In this project, we have looked at the problem of classification of images containing text based on the structural patterns of these scripts.
Project report link.
Project details at Github.
To know more about my work, please click here >>
Project | 01
Project | 02