Fig. 4: Influence of dataset size and data noise on ViT and CNN model performance for printing quality recognition.
From: Transformer-based approach for printing quality recognition in fused filament fabrication
a Performance comparison of ViT and CNN models across varying dataset sizes without artificially introduced noise; b Performance comparison of ViT and CNN models across varying dataset sizes with artificially introduced noise; c Examples of operations applied to raw data to introduce noise.
