GLOBAL WHEAT DETECTION
The data is images of wheat fields, with bounding boxes for each identified wheat head. Not all images include wheat heads / bounding boxes. The images were recorded in many locations around the world.
The CSV data is simple - the image ID matches up with the filename of a given image, and the width and height of the image are included, along with a bounding box. There is a row in train.csv for each bounding box. Not all images have bounding boxes so you have to do labeling yourself and the resolution is 1024x1024 pixels.
Most of the test set images are hidden. A small subset of test images has been included for your use in writing code.
Datset can be downloaded follow the link: https://www.kaggle.com/c/global-wheat-detection/overview
What am I predicting?
You are attempting to predict bounding boxes around each wheat head in images that have them. If there are no wheat heads, you must predict no bounding boxes.
train.csv - the training data
sample_submission.csv - a sample submission file in the correct format
train.zip - training images
test.zip - test images
image_id - the unique image ID
width, height - the width and height of the images
bbox - a bounding box, formatted as a Python-style list of [xmin, ymin, width, height]
The submission format requires a space delimited set of bounding boxes. For example:
ce4833752,0.5 0 0 100 100
indicates that image ce4833752 has a bounding box with a confidence of 0.5, at x == 0 and y == 0, with a width and height of 100.
The file should contain a header and have the following format. Each row in your submission should contain ALL bounding boxes for a given image.
I used old yolo version 2 layers and did not optimize parameter for training. Therefore the accuracy will not be as the expectation. However, I hope this will be useful for everyone aware about MATLAB features and for future optimization.