Update README.md

This dataset is a collection of images of books, intended for machine learning and/or computer vision tasks like image classification. The images have been pre-processed, cropped, and resized, and are categorized into four distinct classes:

• cover-desk: Images of books with their covers visible, placed on a desk.
• cover-floor: Images of books with their covers visible, placed on the floor.
• nocover-desk: Images of books with their covers not visible (e.g., spine), placed on a desk.
• nocover-floor: Images of books with their covers not visible, placed on the floor.

The original images were manually collected and then processed using Python scripts in a Google Colab notebook.

To create the augmented split, extensive data augmentation techniques from the torchvision library were applied. This process involved random resizing, cropping, flipping, rotations, and color jittering to generate many unique variations from the original images. The total number of augmented samples was increased to 300 to improve model training robustness. The final dataset was then pushed to the Hugging Face Hub.

README.md

@@ -13,13 +13,13 @@ dataset_info:
           '3': nocover-floor
   splits:
   - name: original
-    num_bytes: 315420.0
+    num_bytes: 315420
     num_examples: 30
   - name: augmented
-    num_bytes: 21688806.0
+    num_bytes: 21688806
     num_examples: 300
   download_size: 22003966
-  dataset_size: 22004226.0
+  dataset_size: 22004226
 configs:
 - config_name: default
   data_files:
@@ -27,4 +27,16 @@ configs:
     path: data/original-*
   - split: augmented
     path: data/augmented-*
----
+license: mit
+task_categories:
+- image-classification
+language:
+- en
+tags:
+- book_cover
+- Books
+- images
+- classification
+size_categories:
+- n<1K
+---