A Study and Comparison of Human and Deep Learning Recognition Performance Under Visual Distortions
arXiv:1705.02498
Abstract
Deep neural networks (DNNs) achieve excellent performance on standard classification tasks. However, under image quality distortions such as blur and noise, classification accuracy becomes poor. In this work, we compare the performance of DNNs with human subjects on distorted images. We show that, although DNNs perform better than or on par with humans on good quality images, DNN performance is still much lower than human performance on distorted images. We additionally find that there is little correlation in errors between DNNs and human subjects. This could be an indication that the internal representation of images are different between DNNs and the human visual system. These comparisons with human performance could be used to guide future development of more robust DNNs.
References in corpus (6)
- Very Deep Convolutional Networks for Large-Scale Image Recognition
- Going Deeper with Convolutions
- Examining the Impact of Blur on Recognition by Convolutional Networks
- Universal adversarial perturbations
- Quality Resilient Deep Neural Networks
- On Classification of Distorted Images with Deep Convolutional Neural Networks
Cited by in corpus (13)
- Adversarial Examples Are a Natural Consequence of Test Error in Noise
- Benchmarking Neural Network Robustness to Common Corruptions and Surface Variations
- Consistency Regularization for Certified Robustness of Smoothed Classifiers
- On the effectiveness of adversarial training against common corruptions
- On Interaction Between Augmentations and Corruptions in Natural Corruption Robustness
- Parameter-Efficient Transfer from Sequential Behaviors for User Modeling and Recommendation
- Orthogonal Deep Neural Networks
- Towards Corruption-Agnostic Robust Domain Adaptation
- Calibrated neighborhood aware confidence measure for deep metric learning
- R3Net: Random Weights, Rectifier Linear Units and Robustness for Artificial Neural Network
- Natural Perturbed Training for General Robustness of Neural Network Classifiers
- Wiggling Weights to Improve the Robustness of Classifiers
- Generative Sensing: Transforming Unreliable Sensor Data for Reliable Recognition