Summary of Context-aware Multi-model Object Detection For Diversely Heterogeneous Compute Systems, by Justin Davis and Mehmet E. Belviranli
Context-aware Multi-Model Object Detection for Diversely Heterogeneous Compute Systems
by Justin Davis, Mehmet E. Belviranli
First submitted to arxiv on: 12 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper tackles a challenge in deep neural networks (DNNs) used for continuous mobile object detection (OD) tasks, particularly in autonomous systems. The one-size-fits-all approach to DNN deployment results in inefficient utilization of computational resources, which is detrimental in energy-constrained systems. To address this issue, the authors propose SHIFT, a methodology that continuously selects from various DNN-based OD models depending on dynamic contextual information and computational constraints. This selection process considers multi-accelerator execution to optimize energy efficiency while meeting latency requirements. The proposed approach achieves significant improvements of up to 7.5x in energy usage and 2.8x in latency compared to state-of-the-art GPU-based single model OD approaches. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps solve a problem with how we use deep learning for object detection on mobile devices, like those used in self-driving cars. Currently, we just use one type of deep neural network (DNN) for everything, which wastes energy and makes the system slower. The authors suggest a new way to do this called SHIFT, where the computer chooses the right DNN model based on what it’s seeing and how much energy it has left. This helps make the whole system more efficient and faster. |
Keywords
* Artificial intelligence * Deep learning * Neural network * Object detection
