Smoke segmentation using Deep Learning

Developed during the Computer Vision course under the guidance of Dr. Pritee Khanna at IVE Lab, IIITDMJ, this research project uses state-of-the-art deep learning algorithms like Mask RCNN for automatic segmentation of smoke regions in images.

Technologies used: Deep Learning, Computer Vision, Python, Google Colab


Green area represents actual smoke area (ground truth), Red area represents automatic segmented area, Brown region is the overlap between these.