ip102

INSPECT (Image-Text Network for Systematic Pest Classification Task), Multimodal Few-Shot Crop Pest & Disease Recognition with Image-Text Alignment & Fusion.

Code for my Master's Thesis in Data Science: Pest Insect Classification on the IP102 Dataset. The paper focuses on improving classification performance on this dataset by addressing two of its major issues, namely class imbalance and intra-class variance.

About Insect pests significantly impact agricultural yields, and accurate recognition helps prevent economic losses. This project uses transfer learning on the IP102 dataset with ResNet-50 pre-trained on ImageNet.