Scribd
Upload
90 views5 pages

Image Annotation Internship Guide

The document provides instructions for annotating images for person detection using CVAT, including registering for CVAT, downloading a sample dataset, creating a task to label images with bounding boxes for people, heads and hands, exporting the annotations, and sharing the annotated data for internship consideration which will be judged based on quality, quantity and lack of logic errors in under 24 hours.

Uploaded by

Thanvi Uddaraju
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
90 views5 pages

Image Annotation Internship Guide

The document provides instructions for annotating images for person detection using CVAT, including registering for CVAT, downloading a sample dataset, creating a task to label images with bounding boxes for people, heads and hands, exporting the annotations, and sharing the annotated data for internship consideration which will be judged based on quality, quantity and lack of logic errors in under 24 hours.

Uploaded by

Thanvi Uddaraju
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Task Guidelines

Step 1: What is Image Annotation : https://hackernoon.com/what-is-image-annotation-an-intro-to-5-image-annotation-


services-yt6n3xfj?ref=hackernoon.com

Step 2: CVAT Tutorial : How to Label Images for Object Detection with CVAT

Step3: CVAT Tool : https://cvat.org/auth/register Register the CVAT with your email id

Step 4: Download Dataset https://wobotintelligence-


my.sharepoint.com/:f:/g/personal/soudnya_gandhi_wobot_ai/EnD2dqarvZxKp6HJJ28gvigBB5262Aek8SX73pGmdphB
PQ?e=ftjl2W

Step 5: Create new task with task name as your Full Name, and Add labels – Person, Head, Hand and Upload the given
zip file and click submit.
Labels – Hand, Head, Person (you need to draw a rectangular box on the person, hand and head. overlapping of the
boxes is not a problem)

Step 6: Export the annotations in the format: CVAT for images 1.1

Step 7 :Upload the folder on google drive and share at [email protected]

Data Annotation Instructions –Person Detection

Type of Annotation:

• Person Detection
Dataset Labels:

• Person
• Head
• Hand

Application of the Dataset:

• Use-case: Person Detection

Important Annotation Parameters:


• Minimum 30% object should be visible
• Bounding Box Type (For Detections): Tightly Bound / Edge to Edge

Dataset Specific Annotation Instructions:

1. All objects that fall under the classes (person, head and hand) in the image are to be
annotated

2. Heavily Blurred / shaky images like this can be skipped


3. Crowded frames can be skipped for this task like this

4. edge to edge with minimum garbage value

Correct labelling-
Wrong labelling -

5. Always save your work manually using save button every 5 mins or you might lose your work Option for
auto save can be found under setting > Workspace > enable Auto save

If save option doesn’t work kindly reload (u might lose some annotations )
Internship selection criteria will be judged on the bellow points

• Quality of the bounding boxes (refer point 4)


• Quantity of bounding boxes (1000)
• Logic errors Behind annotations can reduce chances of getting internship – highly blur images marked,
bounding box error, etc.

Applicants can submit their submissions after reaching goal of 1000 annotations within 1 day (24-hour period)

To check the annotations number, click on “ i ” button on the top left section of CVAT under Total Tab
To export dataset: Actions (…) > Export as dataset > : CVAT for images 1.1

Share drive link of exported dataset at [email protected]

Recommended for Internship


• Mouse
• Good Internet Connection (daily data usage could reach 1-2 gb)

You might also like