Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Image Classification for CSSVD Detection in Cacao Plants

Atuhurra Jesse, N'guessan Yves-Roland Douha, Pabitra Lenka

TL;DR

The study tackles CSSVD detection in cacao amid limited labeled data by fine-tuning three pretrained architectures—VGG16, ResNet50, and Vision Transformer (ViT)—on the public KaraAgroAI Cocoa Dataset (17,703 images across Anthracnose, CSSVD, and Healthy). It compares CNN-based and transformer-based approaches, reporting that ResNet50 achieves the best performance with 95.39% precision, 93.75% recall, 94.34% F1, and 94% accuracy after only 20 epochs, marking a +9.75 percentage-point recall improvement over prior work. The dataset is split 80/10/10 into train/validation/test sets, with detailed per-class metrics and model sizes (VGG16: ~138.3M, ResNet50: ~23.5M, ViT: ~6.8M). Overall, the results demonstrate that image classifiers can reliably identify CSSVD-infected cacao plants using high-quality public data, supporting scalable disease screening for farmers.

Abstract

The detection of diseases within plants has attracted a lot of attention from computer vision enthusiasts. Despite the progress made to detect diseases in many plants, there remains a research gap to train image classifiers to detect the cacao swollen shoot virus disease or CSSVD for short, pertinent to cacao plants. This gap has mainly been due to the unavailability of high quality labeled training data. Moreover, institutions have been hesitant to share their data related to CSSVD. To fill these gaps, we propose the development of image classifiers to detect CSSVD-infected cacao plants. Our proposed solution is based on VGG16, ResNet50 and Vision Transformer (ViT). We evaluate the classifiers on a recently released and publicly accessible KaraAgroAI Cocoa dataset. Our best image classifier, based on ResNet50, achieves 95.39\% precision, 93.75\% recall, 94.34\% F1-score and 94\% accuracy on only 20 epochs. There is a +9.75\% improvement in recall when compared to previous works. Our results indicate that the image classifiers learn to identify cacao plants infected with CSSVD.

Image Classification for CSSVD Detection in Cacao Plants

TL;DR

The study tackles CSSVD detection in cacao amid limited labeled data by fine-tuning three pretrained architectures—VGG16, ResNet50, and Vision Transformer (ViT)—on the public KaraAgroAI Cocoa Dataset (17,703 images across Anthracnose, CSSVD, and Healthy). It compares CNN-based and transformer-based approaches, reporting that ResNet50 achieves the best performance with 95.39% precision, 93.75% recall, 94.34% F1, and 94% accuracy after only 20 epochs, marking a +9.75 percentage-point recall improvement over prior work. The dataset is split 80/10/10 into train/validation/test sets, with detailed per-class metrics and model sizes (VGG16: ~138.3M, ResNet50: ~23.5M, ViT: ~6.8M). Overall, the results demonstrate that image classifiers can reliably identify CSSVD-infected cacao plants using high-quality public data, supporting scalable disease screening for farmers.

Abstract

The detection of diseases within plants has attracted a lot of attention from computer vision enthusiasts. Despite the progress made to detect diseases in many plants, there remains a research gap to train image classifiers to detect the cacao swollen shoot virus disease or CSSVD for short, pertinent to cacao plants. This gap has mainly been due to the unavailability of high quality labeled training data. Moreover, institutions have been hesitant to share their data related to CSSVD. To fill these gaps, we propose the development of image classifiers to detect CSSVD-infected cacao plants. Our proposed solution is based on VGG16, ResNet50 and Vision Transformer (ViT). We evaluate the classifiers on a recently released and publicly accessible KaraAgroAI Cocoa dataset. Our best image classifier, based on ResNet50, achieves 95.39\% precision, 93.75\% recall, 94.34\% F1-score and 94\% accuracy on only 20 epochs. There is a +9.75\% improvement in recall when compared to previous works. Our results indicate that the image classifiers learn to identify cacao plants infected with CSSVD.
Paper Structure (11 sections, 1 equation, 2 figures, 7 tables)

This paper contains 11 sections, 1 equation, 2 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methods
  4. VGG16
  5. ResNet50
  6. Vision Transformer (ViT)
  7. Experiments
  8. KaraAgroAI Dataset
  9. Experiment Settings
  10. Results and Analysis
  11. Conclusion

Figures (2)

  • Figure 1: ConvNet classifiers for CSSVD detection. Two ConvNets are used in this experiment. That is, VGG16 and ResNet50. In both cases, feature vectors are extracted from images of cacao leaves, stems and fruits. The classifier is then trained to extract distinguishing features to categorize an image into one of three classes, namely: Anthracnose, CSSVD, Healthy.
  • Figure 2: Left: CSSVD prevalence. Right: CSSVD detection with a Vision Transformer.