Optimized Xception Architecture for Edge Devices Using Depthwise Separable and Deep Residual Convolutions

This paper proposes a modified Xception architecture, incorporating depthwise separable convolutions and deep residual connections, to optimize the model for deployment on edge devices while maintaining competitive performance in object detection tasks.