Skip to main content

Datasets

Standard Dataset

Micronutrient Deficiency Data

Citation Author(s):
Amey Agarwal
Shreya Rathod
Riva Rodrigues
Nirmitee Sarode
Dhananjay Kalbande
Submitted by:
Amey Agarwal
Last updated:
DOI:
10.21227/e5pe-qe60
Data Format:
No Ratings Yet

Abstract

Paper : Assessment of Inference Improvements for Facial Micronutrient Deficiency Detection using Attention-Enhanced YOLOv5

Authors : Amey Agarwal, Shreya Rathod, Riva Rodrigues, Nirmitee Sarode, Dhananjay R. Kalbande

Desciption

This is a dataset of 7 classes : 6 facial skin problems and 1 null class.

A facial skin problem may be identified in an image and marked using Bounding Box Annotation.

Acne Class indicates deficiency of Vitamin D

Blackhead and Nodules are types of acne 

Darkspot indicates deficiency of Vitamin B12

Freckle class does not have a reported potential micronutrient deficiency

Skin Redness, treated as hyperpigmentation, indicates deficiency of Vitamin B12

Null class indicates healthy face image

Size

MicronutrientDeficiencyData-1 Folder size : 169 MB

Platform

No specific platform required.

Environmnent

No specific environment required.

Directory

MicronutrientDeficiencyData-1

|--->data.yaml

|--->README.dataset.txt

|--->README.roboflow.txt

|--->test

       |--->images/*.jpg

       |--->labels/*.txt

|--->train

       |--->images/*.jpg

       |--->labels/*.txt

|--->valid

       |--->images/*.jpg

       |--->labels/*.txt

Class Names

class_id : class_name

0 : acne

1 : blackhead

2 : darkspot

3 : freckle

4 : nodules

5 : skinredness

6 : null

TXT file format 

Format : YOLOv5 PyTorch TXT 

class_id center_x center_y width height

Reference Support : https://roboflow.com/formats/yolov5-pytorch-txt

Training Instructions

1) !git clone https://github.com/ultralytics/yolov5

2) !pip install -r yolov5/requirements.txt

3) !python yolov5/train.py --img 640 --batch 16 --epochs 300 --data yolov5/MicronutrientDeficiencyData-1/data.yaml --cfg yolov5/models/yolov5s.yaml --weights '' --save-period 10

References (Open-Source Datasets)

1) https://universe.roboflow.com/testing-rfihx/facial-health/dataset/1

2) https://github.com/xpwu95/LDL

3) https://universe.roboflow.com/kritsakorn/acne-kbm0q

4) https://universe.roboflow.com/bdle-9kwmc/skin-problems-2eamt

5) https://universe.roboflow.com/dermafie/dermafie

6) https://universe.roboflow.com/sanjanar-yktgw/normal-bwevz

Author Contact

Amey Agarwal : amey.agarwal@spit.ac.in

Dhananjay R. Kalbande : drkalbande@spit.ac.in

Shreya Rathod : shreya.rathod22@spit.ac.in

Nirmitee Sarode : nirmitee.sarode22@spit.ac.in

Riva Rodrigues : riva.rodrigues22@spit.ac.in

 

Instructions:

1) !git clone https://github.com/ultralytics/yolov5

2) !pip install -r yolov5/requirements.txt

3) !python yolov5/train.py --img 640 --batch 16 --epochs 300 --data yolov5/MicronutrientDeficiencyData-1/data.yaml --cfg yolov5/models/yolov5s.yaml --weights '' --save-period 10

This is the cleaned preprocesed dataset for detecting facial skin issues(bounding box) to identify potential micronutrient deficiencies

Amey Agarwal Wed, 04/16/2025 - 11:05 Permalink