Modular Transformer Architecture for Precision Agriculture Imaging

2025-08-04

AI Summary

To address the accuracy-efficiency trade-off in weed segmentation from drone-captured agricultural videos—exacerbated by image degradations such as blur and noise—this paper proposes a quality-aware modular Vision Transformer (ViT) framework. It introduces a lightweight, novel image quality assessment mechanism based on Mean Absolute Deviation and the Laplacian operator, enabling dynamic routing to three specialized submodels: a baseline ViT, a Fisher Vector-enhanced ViT, and a Lucy-Richardson deconvolution–integrated ViT decoder. Quality-adaptive preprocessing and architecture customization significantly improve robustness and segmentation accuracy. Experiments demonstrate that our method achieves a 6.2% higher mean Intersection-over-Union (mIoU) for weed segmentation compared to state-of-the-art CNNs, while accelerating inference by 23%. The framework thus delivers both high precision and low computational overhead, enabling real-time, accurate weed identification in complex field environments.

Technology Category

• Computer Vision: Segmentation
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Hardware-aware ML

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines

Abstract

This paper addresses the critical need for efficient and accurate weed segmentation from drone video in precision agriculture. A quality-aware modular deep-learning framework is proposed that addresses common image degradation by analyzing quality conditions-such as blur and noise-and routing inputs through specialized pre-processing and transformer models optimized for each degradation type. The system first analyzes drone images for noise and blur using Mean Absolute Deviation and the Laplacian. Data is then dynamically routed to one of three vision transformer models: a baseline for clean images, a modified transformer with Fisher Vector encoding for noise reduction, or another with an unrolled Lucy-Robinson decoder to correct blur. This novel routing strategy allows the system to outperform existing CNN-based methods in both segmentation quality and computational efficiency, demonstrating a significant advancement in deep-learning applications for agriculture.

Problem

• Efficient weed segmentation from drone video in agriculture
• Handling image degradation like blur and noise effectively
• Improving segmentation quality and computational efficiency over CNNs

Innovation

• Modular deep-learning framework for quality-aware processing
• Dynamic routing to specialized transformer models
• Fisher Vector and Lucy-Robinson decoders for degradation