A Kaggle-hosted dataset of plant disease images formatted for computer vision models. The collection contains over 38,000 images, is balanced and augmented, and each image is sized at 256×256 pixels. The dataset is specifically prepared for training and evaluating the EfficientNet-B4 architecture.
Use Cases
- Training image classification models for crop disease detection based on the 256×256 image format.
- Benchmarking the performance of EfficientNet-B4 and similar architectures on agricultural imagery.
- Developing automated plant health monitoring systems based on the balanced and augmented dataset.
- Creating educational or demonstration tools for agricultural AI based on the labeled disease images.
Strengths
- Contains over 38,000 images, providing a substantial volume for model training.
- Images are pre-processed to a uniform 256×256 pixel size, simplifying input pipelines.
- The dataset is described as balanced and augmented, which can improve model robustness.
Limitations
- Description metadata is limited; actual data quality requires manual inspection after download.
- Column-level documentation is absent; field semantics must be inferred after download.
- Data may reflect geographic or source bias inherent to Kaggle-hosted collections.