Kamau-Johnson/Exploring-Data_Sets-Crop-Disease-in-Uganda-

Explored data_sets in Crop Disease in Uganda Project ## 2.2 Explore Dataset Warning! Content Updated! This notebook has been update since the original creation and recording of the video. Originally, this notebook pulled data from GCP, but we've made a change so it's always available without the need of pulling it. Our data for this project is related to a Kaggle competition about crops in Uganda. Here, we'll work with images of cassava plants, an extremely important staple crop. We have images of both healthy plants, and plants with one of four diseases. The goal of the competition is to create a model to diagnose what disease a plant has, or if it's healthy. This would be the first step toward speeding up this diagnosis to help farmers maintain their crops. We'll specifically be looking at images from this Kaggle Dataset, which has data from the 2019 and 2020 versions of this competition. **Content Update Warning** Changes with respect to the videos You may notice that the number of files in the category directories, located inside data_p2/train/ , is much lower than what is shown in the video. The video uses the full dataset with thousands of images per category (e.g., 2863 images in cassava-healthy ). For performance reasons, this lab has been updated to use a smaller, representative subset of the data (e.g., 38 images in cassava-healthy ). This change allows the lab to run much faster. The steps you need to follow and the concepts being taught are not affected by this change. The subsequent labs contain the right number of files. The data is available at the location `data_p2/train`. You can explore the dataset freely and then move to the next section. ```python !ls -l data_p2 ``` total 24264 drwxr-xr-x 206 root root 24846336 Jan 9 10:55 train **Task 2.2.1:** Create a variable for the `train` directory. ```python data_dir = "data_p2" train_dir = os.path.join(data_dir, "train") print("Data directory:", train_dir) ``` Data directory: data_p2/train In the training directory, each class …