Energy-Efficient Uncertainty-Aware Biomass Composition Prediction at the Edge

TL;DR

The paper addresses energy-limited edge deployment of biomass composition prediction from field images by combining energy-aware filter pruning at initialization with an uncertainty-aware regression framework. A variance attenuation loss trains pruned models to output a predictive distribution $y \sim \mathcal{N}(\mu(x), \sigma(x))$, enabling on-device uncertainty assessment. When predictions are uncertain ($\sigma$ high), the system re-infers with an unpruned model, creating a hybrid that reduces energy use while preserving accuracy. Evaluations on GrassClover and Irish Clover datasets on an NVIDIA Jetson Nano show about $50\%$ energy reduction with only around $4\%$ RMSE loss, demonstrating practical edge deployment for targeted fertilization and local sowing decisions.

Abstract

Clover fixates nitrogen from the atmosphere to the ground, making grass-clover mixtures highly desirable to reduce external nitrogen fertilization. Herbage containing clover additionally promotes higher food intake, resulting in higher milk production. Herbage probing however remains largely unused as it requires a time-intensive manual laboratory analysis. Without this information, farmers are unable to perform localized clover sowing or take targeted fertilization decisions. Deep learning algorithms have been proposed with the goal to estimate the dry biomass composition from images of the grass directly in the fields. The energy-intensive nature of deep learning however limits deployment to practical edge devices such as smartphones. This paper proposes to fill this gap by applying filter pruning to reduce the energy requirement of existing deep learning solutions. We report that although pruned networks are accurate on controlled, high-quality images of the grass, they struggle to generalize to real-world smartphone images that are blurry or taken from challenging angles. We address this challenge by training filter-pruned models using a variance attenuation loss so they can predict the uncertainty of their predictions. When the uncertainty exceeds a threshold, we re-infer using a more accurate unpruned model. This hybrid approach allows us to reduce energy consumption while retaining a high accuracy. We evaluate our algorithm on two datasets: the GrassClover and the Irish clover using an NVIDIA Jetson Nano edge device. We find that we reduce energy reduction with respect to state-of-the-art solutions by 50% on average with only 4% accuracy loss.

Figures (8)

• Figure 1: A typical workflow of deep learning based biomass estimation from phone / camera images.
• Figure 2: Our proposed framework based on variance-guided energy efficient deep learning for biomass estimation on edge.
• Figure 3: Examples of predicted high and low variance images from the Irish dataset, ResNet18 $80\%$ pruned.
• Figure 4: Examples of predicted high and low variance images from the GrassClover dataset, ResNet18 $80\%$ pruned.
• Figure 5: (a) showing the correlation between RMSE and Variance, (b) showing the RMSE distribution over test set of Irish clover dataset for VGG-16.