Scribd
Upload
34 views

Internship Assignment CV

The internship assignment asks interns to develop a solution using image processing techniques to partially de-annotate an image by removing specific annotations from a fully annotated image while preserving the underlying information from the original image. The solution must be able to generalize to different types of images and annotations and process large datasets efficiently. It will be evaluated on sample images containing original, fully annotated, and partially annotated images. The submission requirements include a Python script with a function to generate a partially annotated output image from the original and fully annotated images along with documentation and adherence to coding standards.

Uploaded by

Shruthy Menon
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
34 views

Internship Assignment CV

The internship assignment asks interns to develop a solution using image processing techniques to partially de-annotate an image by removing specific annotations from a fully annotated image while preserving the underlying information from the original image. The solution must be able to generalize to different types of images and annotations and process large datasets efficiently. It will be evaluated on sample images containing original, fully annotated, and partially annotated images. The submission requirements include a Python script with a function to generate a partially annotated output image from the original and fully annotated images along with documentation and adherence to coding standards.

Uploaded by

Shruthy Menon
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

Internship Assignment

for Computer Vision


Creating a partially annotated image using
fully annotated and a reference image

11th April 2023


Objective
The objective of this assignment is to develop a solution using image processing techniques
to partially de-annotate an image, given an original image and a fully annotated image. The
solution should be able to de-annotate the image partially without re-annotating the original
image. The assignment will also provide a set of sample images containing original, fully
annotated, and partially annotated images for testing and evaluation purposes.The solution
should be able to identify and remove specific annotations from the fully annotated image
while preserving the underlying information of the original image. The partially de-annotated
image should not contain any unwanted annotations or distortions that may affect its
interpretation. Additionally, the solution should be able to generalize to different types of
images and annotations, and should be efficient and scalable for processing large datasets.

** Annotation of original images partially will not be considered as a solution.


You’re not allowed to use any Object Detection model on the images

Data
Please download the assignment folder from here. The file will contain 2 folder train and
test and a submission file.
The train folder contains 3 images
● original_image : Original Base Image without annotation
● fully_annotated : Fully annotated Image
● partially_annotated : Partially annotated Image

The test folder will have 2 images.


● original_image : Original Base Image without annotation
● fully_annotated : Fully annotated Image

Assumptions
You can assume the following in the image:
● There will be only two annotations per image
● There will be one dog and a cat in the image
● The annotation of the cat should be retained in the partially annotated image

akaike.ai 1
Example Images

Submissions
The Submission shall be output image saved in output image path

Code format
The code submission shall be a .py script that contains the following function. Each function
is associated with an objective as explained in the objective section.

def get_ouput_image(original_image_path: str, fully_annotated_image_path: str,


partially_annotated_image_path: str):

cv2.imwrite(partially_annotated_image_path, partially_annotated_image)

The above functions shall have the following behaviour:


1. The functions shall only accept arguments of string data types and list. If any other
data type is passed, it should raise an exception.
2. The functions should not return anything

akaike.ai 2
Submission guidelines
1. Your code and submission will be subject to automated checks, unit tests and
plagiarism checks. Please follow the below guidelines so that it is not rejected.
2. Your code will also be automatically checked for code quality as well. Please follow
PEP8 guidelines.
3. Share a writeup discussing the methods and techniques you’ve used in creating your
solution.
4. Share the submission as a script (.py) file and include proper documentation that
supports your solution.
5. Zip your code and documentation with the following naming convection and upload to
the appropriate slot in the google forms <yourname>_cv_assignment.zip

Solution Guidelines
1. Feel free to use any image processing techniques, libraries. Please provide due
credits to those.
2. Try OpenCV based techniques.
3. Do not use any Object Detection Models for Annotation. Using one would void your
submission

akaike.ai 3

You might also like