Efficient Deep Learning Framework for ۳D Printer Failure Detection Using a Lightweight Mobile ViT-Inspired Network

Additive manufacturing systems, particularly fused deposition modeling (FDM) ۳D printers, are prone to print failures that lead to material waste, time loss, and reduced production reliability. This study proposes a lightweight deep learning approach for automatic detection of ۳D printing failures using an image-based classification framework. A custom Mini Mobile ViT-inspired convolutional architecture is designed to provide an efficient balance between computational cost and predictive accuracy, making it suitable for deployment in resource-constrained environments. The model is trained on a publicly available image dataset containing normal and failed print samples, with standard preprocessing and normalization applied to ensure consistent input representation. Experimental evaluation demonstrates that the proposed network successfully learns discriminative visual features associated with common printing defects while maintaining a compact parameter footprint. The training results indicate stable convergence and promising classification performance despite the model's simplicity. The proposed method shows potential for real-time monitoring of ۳D printing processes and offers a practical foundation for embedded quality-control systems in smart manufacturing pipelines.

کلیدواژه ها:

۳D printing ، failure detection ، Mobile ViT ، lightweight deep learning ، computer vision ، additive manufacturing