Lightweight Deep Learning Model for Wireless Capsule Endoscopy Image Classification Using SqueezeNet

Wireless Capsule Endoscopy has revolutionized gastrointestinal imaging by furnishing a non-invasive system to capture images of the entire digestive tract. This paper proposes a lightweight deep learning model for WCE image using the SqueezeNet enhanced with an attention mechanism. SqueezeNet, known for its compact armature, significantly reduces the number of parameters without compromising performance, making it suitable for resource - constrained surroundings. By integrating an attention medium, the model widely focuses on the most appropriate features in the WCE images. Experimentation has been carried out over a dataset of gastrointestinal images, and the proposed model achieved greater classification accuracy compared to traditional deep learning models, with significantly reduced computational cost. The lightweight nature of this model makes it ideal for deployment in real- time clinical operations.