Intelligent generation and editing optimization of digital media image content based on Deep Learning and the Internet of Things

The intelligent generation and editing optimization of digital media image content have become vital in the era of interconnected smart systems. This research proposes a comprehensive framework that integrates Deep Learning (DL) and the Internet of Things (IoT) to enable automated, context-aware generation and adaptive editing of image content. The framework utilizes data collected from IoT-enabled imaging devices, which capture contextual attributes. The collected Low-Light Image Enhancement Dataset ( n = 970) paired images for enhancement and restoration research involved preprocessing through normalization and Contrast Limited Adaptive Histogram Equalization (CLAHE) to improve difference, visibility, ensures high-quality input for model training. In the feature extraction stage, a hybrid Convolutional Neural Network (CNN) and Vision Transformer (ViT) architecture is employed to identify semantic features, object boundaries, and contextual relationships across multiple image regions. The extracted feature maps are then used within a Glowworm Swarm mutated Attention-based Generative Adversarial Network (GWS-Att-GAN) for optimized image generation and editing. Glowworm Swarm (GWS) efficiently explores optimal feature mappings, while the attention mechanism focuses on key image regions to improve realism and contextual accuracy. This GAN also incorporates an attention-guided refinement module to preserve essential regions and enhance visual quality. Experimental results indicate that the proposed GWS-Att-GAN implemented using Python 3.10 that performs better, with PSNR of 34.2 dB, SSIM of 0.93, and diversity score of 0.88. Additionally, it improves contextual adaptation efficiency (62.3%) and IoT synchronization reliability (88.6%). When contrasted with conventional Att-GAN, the model shows improved convergence stability and enhanced context-aware picture generation capacities.