Efficient neural network backbones for mobile devices are often optimized for metrics such as FLOPs or parameter count. However, these metrics may not correlate well with latency of the network when deployed on a mobile device. Therefore, we perform extensive analysis of different metrics by deploying several mobile-friendly networks on a mobile device. We identify and analyze architectural and optimization bottlenecks in recent efficient neural networks and provide ways to mitigate these bottlenecks. To this end, we design an efficient backbone MobileOne, with variants achieving an inference time under 1 ms on an iPhone12 with 75.9% top-1 accuracy on ImageNet. We show that MobileOne achieves state-of-the-art performance within the efficient architectures while being many times faster on mobile. Our best model obtains similar performance on ImageNet as MobileFormer while being 38× faster. Our model obtains 2.3% better top-1 accuracy on ImageNet than EfficientNet at similar latency. Furthermore, we show that our model generalizes to multiple tasks – image classification, object detection, and semantic segmentation with significant improvements in latency and accuracy as compared to existing efficient architectures when deployed on a mobile device.

Related readings and updates.

Modeling the Impact of User Mobility on COVID-19 Infection Rates Over Time

As the COVID-19 pandemic took off during early 2020, widespread interest in modeling the trajectory of infections emerged. This interest was predicated on the hope that accurate models could be developed and subsequently used to help governments and policy makers monitor the effect of lockdowns and determine safe points in time to reopen.

See article details

Self-supervised Semi-supervised Learning for Data Labeling and Quality Evaluation

This paper was accepted at the Data-Centric AI Workshop at the NeurIPS 2021 conference. As the adoption of deep learning techniques in industrial applications grows with increasing speed and scale, successful deployment of deep learning models often hinges on the availability, volume, and quality of annotated data. In this paper, we tackle the problems of efficient data labeling and annotation verification under the human-in-the-loop setting. We…
See paper details