Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Evaluating the Energy Efficiency of Few-Shot Learning for Object Detection in Industrial Settings

Georgios Tsoumplekas, Vladislav Li, Ilias Siniosoglou, Vasileios Argyriou, Sotirios K. Goudos, Ioannis D. Moscholios, Panagiotis Radoglou-Grammatikis, Panagiotis Sarigiannidis

TL;DR

A novel way to quantify the trade-off between performance and efficiency in model training and energy consumption is introduced through a customized Efficiency Factor metric.

Abstract

In the ever-evolving era of Artificial Intelligence (AI), model performance has constituted a key metric driving innovation, leading to an exponential growth in model size and complexity. However, sustainability and energy efficiency have been critical requirements during deployment in contemporary industrial settings, necessitating the use of data-efficient approaches such as few-shot learning. In this paper, to alleviate the burden of lengthy model training and minimize energy consumption, a finetuning approach to adapt standard object detection models to downstream tasks is examined. Subsequently, a thorough case study and evaluation of the energy demands of the developed models, applied in object detection benchmark datasets from volatile industrial environments is presented. Specifically, different finetuning strategies as well as utilization of ancillary evaluation data during training are examined, and the trade-off between performance and efficiency is highlighted in this low-data regime. Finally, this paper introduces a novel way to quantify this trade-off through a customized Efficiency Factor metric.

Evaluating the Energy Efficiency of Few-Shot Learning for Object Detection in Industrial Settings

TL;DR

A novel way to quantify the trade-off between performance and efficiency in model training and energy consumption is introduced through a customized Efficiency Factor metric.

Abstract

In the ever-evolving era of Artificial Intelligence (AI), model performance has constituted a key metric driving innovation, leading to an exponential growth in model size and complexity. However, sustainability and energy efficiency have been critical requirements during deployment in contemporary industrial settings, necessitating the use of data-efficient approaches such as few-shot learning. In this paper, to alleviate the burden of lengthy model training and minimize energy consumption, a finetuning approach to adapt standard object detection models to downstream tasks is examined. Subsequently, a thorough case study and evaluation of the energy demands of the developed models, applied in object detection benchmark datasets from volatile industrial environments is presented. Specifically, different finetuning strategies as well as utilization of ancillary evaluation data during training are examined, and the trade-off between performance and efficiency is highlighted in this low-data regime. Finally, this paper introduces a novel way to quantify this trade-off through a customized Efficiency Factor metric.
Paper Structure (15 sections, 3 equations, 6 figures, 4 tables)

This paper contains 15 sections, 3 equations, 6 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Object Detection
  4. Few-Shot Object Detection
  5. AI Model Energy Efficiency
  6. Methodology
  7. Model Architecture
  8. Few-shot Object Detection
  9. Few-Shot Learning via Model Finetuning
  10. Experimental Results
  11. Experimental Configuration
  12. Datasets
  13. Evaluation Metrics
  14. Experimental Results
  15. Conclusions

Figures (6)

  • Figure 1: Illustration of the two-step training procedure based on base class pretraining and novel class finetuning.
  • Figure 2: Dataset Preview. (a) PPE, (b) Construction Safety, (c) Fire Detection
  • Figure 3: Mean mAP and standard deviation of $best$ models for varying finetuning strategies and numbers of shots.
  • Figure 4: mAP with respect to energy consumption for different finetuning strategies and numbers of shots.
  • Figure 5: Model performance with respect to energy consumption in the PPE dataset.
  • ...and 1 more figures