The MangoNet Semantic Dataset

The MangoNet semantic dataset contains images of mango trees in an open orchad and their correspoding semantic segmented images. The MangoNet semantic dataset has been used to develop a deep learning framework to automatically detect and count mangoes in an open orchad.

Data description

The MangoNet semantic dataset contains 45 training images of size 4000 x 3000 and contains 4 test images of size 4000 x 3000. For each image the dataset contains a correspoding annotated image which is colored green in regions of mangoes and black in non mango regions.

Salient features of Dataset:

1. Contains images with varied number of mangoes.
2. Contains images with mangoes in different scale and size.
3. Contains images with varying contrast and lighting.
4. Contains images with mangoes present in vayinng amount of occlusion.

Example images of the MangoNet Semantic Dataset:

Original Images	Annotated Images

Cloning the MangoNet semantic dataset.

Clone the dataset from git using:

git clone https://github.com/avadesh02/MangoNet-Semantic-Dataset.git

How to cite the MangoNet semantic dataset ?

Please cite the dataset in your publication if it helps your research.

Title of paper : MangoNet:A deep semantic segmentation architecture for a method to detect and count mangoes in an open orchard

Status : Paper under review in International Scientific Journal Engineering Applications of Artificial Intelligence.