Weed Recognition using Deep Learning Techniques on Class-imbalanced Imagery

TL;DR

This study evaluates five deep neural networks for weed recognition in agricultural imagery, addressing class imbalance through data augmentation and transfer learning, and finds different models excel depending on dataset size.

Contribution

It compares multiple deep learning models for weed recognition, introduces a large combined dataset, and explores transfer learning to improve accuracy in imbalanced conditions.

Findings

VGG16 outperforms others on small datasets

ResNet-50 performs best on large combined dataset

Data augmentation and transfer learning enhance recognition accuracy

Abstract

Most weed species can adversely impact agricultural productivity by competing for nutrients required by high-value crops. Manual weeding is not practical for large cropping areas. Many studies have been undertaken to develop automatic weed management systems for agricultural crops. In this process, one of the major tasks is to recognise the weeds from images. However, weed recognition is a challenging task. It is because weed and crop plants can be similar in colour, texture and shape which can be exacerbated further by the imaging conditions, geographic or weather conditions when the images are recorded. Advanced machine learning techniques can be used to recognise weeds from imagery. In this paper, we have investigated five state-of-the-art deep neural networks, namely VGG16, ResNet-50, Inception-V3, Inception-ResNet-v2 and MobileNetV2, and evaluated their performance for weed recognition. We have used several experimental settings and multiple dataset combinations. In particular, we constructed a large weed-crop dataset by combining several smaller datasets, mitigating class imbalance by data augmentation, and using this dataset in benchmarking the deep neural networks. We investigated the use of transfer learning techniques by preserving the pre-trained weights for extracting the features and fine-tuning them using the images of crop and weed datasets. We found that VGG16 performed better than others on small-scale datasets, while ResNet-50 performed better than other deep networks on the large combined dataset.