semantic segmentation

SELECTIVE ENCRYPTION OF VIDEO INFORMATION BASED ON SEMANTIC SEGMENTATION USING U-NET NEURAL NETWORK

In the digital age, video information has taken a leading place among data types in terms of volume and significance. Large amounts of visual data are created daily using video cameras, mobile gadgets, drones and network services, and a significant part of this content may contain personal or confidential information. Although traditional full encryption of the video stream guarantees a high level of protection, it is accompanied by a number of disadvantages: high load on computing resources, delays during data transmission and difficulties in implementing real-time processing.

Improving pedestrian segmentation using region proposal-based CNN semantic segmentation

Pedestrian segmentation is a critical task in computer vision, but it can be challenging for segmentation models to accurately classify pedestrians in images with challenging backgrounds and luminosity changes, as well as occlusions.  This challenge is further compounded for compressed models that were designed to deal with the high computational demands of deep neural networks.  To address these challenges, we propose a novel approach that integrates a region proposal-based framework into the segmentation process.  To evaluate the performance of the proposed framework,