MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications

Authors: Howard, A. G., Zhu, M., Chen, B., et al. Year: 2017 Venue: CVPR ArXiv: https://arxiv.org/abs/1704.04861

Summary

MobileNet introduces depthwise separable convolutions for efficient mobile-friendly CNNs. It achieves competitive accuracy with minimal parameters, enabling on-device inference.

Key Concepts

• Depthwise separable convolutions
• Parameter reduction techniques
• Mobile-optimized architecture
• Width and resolution multipliers
• Efficient feature extraction

Impact

MobileNet revolutionized on-device ML by proving that small networks could be highly efficient without sacrificing accuracy. It inspired MobileNetV2, MobileNetV3, and edge ML.

Related Papers

• ShuffleNet: Efficient Architecture via Channel Shuffle
• SqueezeNet: AlexNet-level Accuracy
• EfficientNet: Scaling CNNs Efficiently

---

Citation:

@inproceedings{howard2017mobilenets,
  title={MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications},
  author={Howard, Andrew G and Zhu, Menglong and Chen, Bo and Kalenichenko, Dmitry and others},
  booktitle={IEEE Conference on Computer Vision and Pattern Recognition},
  year={2017}
}