We propose MedNet, a lightweight and efficient CNN architecture that integrates depthwise separable convolutions and CBAM to improve classification accuracy on medical image datasets. We design a ...