Fire Detection using CCTV images

Goal

• Make a fire-detection deep learning model using the Monk library and transfer learning.
• Make a webapp to detect fire in a image.

About the Dataset.

• As the dataset of CCTV images were very less. Therefore I took 2 datasets from Kaggle.
• The first one contained images of fire and non-fire, and the second one contained CCTV images of fire and non-fire.
• Then I merged these two datasets for greater accuracy.

Table of contents in Jupyter Notebook.

• Install Monk
• Using the pre-trained model for fire detection
• Training a classifier from scratch
  • Loading the Dataset
  • Importing Keras backend
  • Training model using mobilenet_v2 as transfer learning architecture
  • Training model using desnenet121 as transfer learning architecture
  • Training model using densenet201 as transfer learning architecture
  • Comparing all the models.
• Conclusion

Visit the blog for more details.

Make web application using Flask library.