BDRice: A comprehensive image dataset of bangladeshi rice classification in computer vision applications
Description
This dataset introduces a newly developed dataset named "BDRice" which includes eight different types of rice from various regions in Dhaka city such as Uttara and Mirpur in Bangladesh. The eight different rice species are formerly defined with their local names. These are Mozumder , Mozzammel-Miniket, Utsob Nazir, Mizan ,Haski ,Pyzam, Nabil ,Atash. The dataset contains a total of 2,756 original images, filtered images 920 along with an additional 3,680 augmented images. We meticulously captured each original image under natural lighting conditions with an appropriate background. This comprehensive dataset holds immense potential for researchers in utilizing various ML and DL methods to advance the healthcare sector significantly. It is a valuable resource for future investigations, laying the foundation for crucial developments in these domains.
Files
Steps to reproduce
To collect the image dataset, the following steps will be considered: Step 1. Rice Selection: Selected rice that is easily found in the local area.The set contains a total of 8 distinct classes of Rice.The classes are 1.Haski 2. Nabil 3. Mozumder 4. Utsob Nazir 5. Mizan 6. Atash 7. Pyzam 8. Mozammel Miniket. Step 2. Image Capturing: Capturing an image of selected rice in two verities: indoors (with a visible surface) and outdoors (that can usually be seen). For clicking the images, the devices used are Canon 700D (18MP), One Plus 8T (1080∗2400 pixels, 48 MP), Samsung F23(1080 x 2408 pixels, 50MP), Samsung Galaxy S8(1440 x 2960 pixels, 12MP), Samsung A55(1080 x 2340 pixels, 50MP). The picture was collected from the shop of the local areas of Uttara, Mirpur in Dhaka. Step 3. Data Storing: Data is stored in a specific file for future use. The dataset contains a total of 3680 images with 460 in each class. Step 4. Data Cleaning: Clean the data by removing blurred, noisy, invisible, and dark images from stored data. Step 5. Data Augmentation: Processing the data for representing a formative dataset and adjusting the image's size, angle, shape, and brightness. Step 6. Final dataset: Store the output obtained from augmentation in an understandable and clear dataset.