Quantization Effects on Neural Networks Perception: How would quantization change the perceptual field of vision models?
This work addresses the underexplored impact of quantization on model perceptual fields, providing insights for deploying efficient and interpretable vision models on resource-limited devices, though it is incremental in nature.
This study investigated how quantization affects the spatial recognition abilities of vision models by analyzing changes in class activation maps (CAMs) and their alignment with salient object maps across six CNN architectures on 10,000 ImageNet images. The results showed differing sensitivities to quantization, impacting model performance and interpretability for real-world deployment.
Neural network quantization is a critical technique for deploying models on resource-limited devices. Despite its widespread use, the impact of quantization on model perceptual fields, particularly in relation to class activation maps (CAMs), remains underexplored. This study investigates how quantization influences the spatial recognition abilities of vision models by examining the alignment between CAMs and visual salient objects maps across various architectures. Utilizing a dataset of 10,000 images from ImageNet, we conduct a comprehensive evaluation of six diverse CNN architectures: VGG16, ResNet50, EfficientNet, MobileNet, SqueezeNet, and DenseNet. Through the systematic application of quantization techniques, we identify subtle changes in CAMs and their alignment with Salient object maps. Our results demonstrate the differing sensitivities of these architectures to quantization and highlight its implications for model performance and interpretability in real-world applications. This work primarily contributes to a deeper understanding of neural network quantization, offering insights essential for deploying efficient and interpretable models in practical settings.