Detecting and Classifying Inappropriate Content in Youtube Videos Using Deep Learning Approach
Abstract: The proliferation of hate speech, pornography, and violence in online platforms is a significant concern, especially in video content on platforms like YouTube. An automated solution can help identify and remove such content, creating a safer and more positive online environment. The main objective of this project is to identify and classify undesirable content in YouTube videos using a variety of deep learning approaches. The suggested method analyzes video frames and audio segments using Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks. Creating a dataset of YouTube videos, pre-processing them to extract pertinent visual and audio attributes, and then training the CNN-LSTM model to discover the spatial and temporal relationships between the video frames and audio segments are all steps in the procedure. On a test set of YouTube videos that have been flagged as unsuitable or not by human annotators, Using measurements of accuracy, precision, recall, and F1-score, the model's performance will be evaluated.
Keywords: Detection, Classification, Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks
How to Cite?: Sanaboina Chandra Sekhar, Yandamuri Eswara Anil, "Detecting and Classifying Inappropriate Content in Youtube Videos Using Deep Learning Approach", Volume 12 Issue 9, September 2023, International Journal of Science and Research (IJSR), Pages: 1447-1451, https://www.ijsr.net/getabstract.php?paperid=MR23914100015, DOI: https://dx.doi.org/10.21275/MR23914100015