Medicinal Plant Classification using Deep Learning Techniques

Published:



  • Our project focuses on classifying medicinal plant leaves based on shape and size using deep learning approaches.
  • We curated a dataset consisting of 30 classes, each containing 50 images. To enhance model generalization, we applied 16 types of augmentations to introduce variety from real-world camera-captured images.
  • We fine-tuned six deep CNN-based models: ResNet50, InceptionV3, VGG19, Xception, EfficientNetV2L, and NASNetLarge.
  • To provide explainability, we used GRADCAM to visualize the regions of the input images influencing the models’ classification decisions

Github: Link