search

abhanjac / blood_cell_image_segmentation

This project shows an automated deep learning based computer vision algorithm that can provide a complete blood cell count (i.e. the number red blood cells, number of white blood cells, different types of white blood cells, number of plateletes etc.) in a person's body from a digitized blood smear image of his blood sample. The neural network used here is a modified version of the Unet framework that identifies different type of blood cells and semantically segments them creating the output image where the different cell regions are indicated with different colors. Additionally, it can also detect the presence of malarial pathogens in the blood. The final test accuracy obtained by the network was 93%.
16 stars 3 forks source link

Dataset sharing #1

Open biototem opened 3 years ago

biototem commented 3 years ago

Is it possible to share the dataset with masks through googledrive links? thx

abhanjac commented 3 years ago

Hi,

I did this project as a part of my Internship at Intel last year (2019), so the dataset is their proprietary asset. I no longer have access to it any more. Its a closed dataset that basically I created for them. But I think I have mentioned the datasets I used to create the training images. You can make use of those to make it yourself. Its not that difficult actually.

Thanks.

Arindam B. Chowdhury

Ph.D. student,

Purdue University.

From: biototem notifications@github.com Sent: Sunday, September 20, 2020 10:44 AM To: abhanjac/blood_cell_image_segmentation blood_cell_image_segmentation@noreply.github.com Cc: Subscribed subscribed@noreply.github.com Subject: [abhanjac/blood_cell_image_segmentation] Dataset sharing (#1)

Is it possible to share the dataset with masks through googledrive links? thx

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/abhanjac/blood_cell_image_segmentation/issues/1, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AHQOYEJC7JRFFWHEW2ZX2KTSGYILHANCNFSM4RTVQLOA.