Download as pdf or txt
Download as pdf or txt
You are on page 1of 2

Milestone Report 2

Project: Traffic Signs Recognition

GROUP: G

TEAM MEMBERS:
Jasleen Kaur (C0886477)
Komal Astawala (C0883392)
Mayankkumar Patel (C0883127)
Shreeprada Devaraju (C0887257)
Simran (C0896133)

Packages used:
1. cv2 (OpenCV): OpenCV is used for image processing tasks like loading, manipulating, and
analyzing images. In traffic sign recognition, it's vital for tasks such as image preprocessing,
feature extraction, and object detection due to its extensive library of computer vision
algorithms.

2. NumPy: NumPy is used for numerical computations and array operations. In traffic sign
recognition, NumPy helps with efficient handling and manipulation of image data arrays, making
tasks like data preprocessing, feature extraction, and model training more streamlined and
computationally efficient.

3. os: The os module provides functions for interacting with the operating system. In traffic sign
recognition, it's used for tasks like file handling, directory navigation, and managing paths to
access datasets, store trained models, or organize processed images.

4. random: The random module provides functions for generating random numbers. In traffic
sign recognition, randomness might be employed in tasks such as data augmentation (e.g.,
randomly rotating or scaling images) to increase the diversity of training data and improve the
robustness of the recognition system.

Augmentation Techniques Used:


1. Rotate Image: Rotate the input image by a specified angle around its center, which can help
increase the variety of training data and improve the model's ability to recognize signs from
different perspectives.
2. Flip Image: Flip the input image horizontally or vertically, simulating different orientations of
the traffic signs. This augmentation technique enhances the model's robustness by providing
additional variations in the dataset.

3. Translate Image: Shift the image horizontally and vertically, mimicking changes in the position
of the traffic signs within the frame. This augmentation helps diversify the dataset and improve
the model's ability to recognize signs at different positions in the image.

4. Adjust Brightness: Alter the brightness of the image by scaling pixel values, which can
simulate variations in lighting conditions. This technique ensures that the model can recognize
signs under different lighting environments.

5. Shear Image: Apply a shear transformation to the image, distorting it along one axis, which
can simulate perspective changes. This augmentation technique introduces variations in the
dataset, making the model more robust to different viewing angles.

6. Grayscale Image: Convert the image to grayscale, reducing the number of color channels.
Grayscale images simplify the input data while retaining essential features, making it easier for
the model to focus on relevant information during training.

Here we iterate through all the 43 folders to get the existing number of images, If the folder
already has 500 or more existing images, no augmentation is done. Otherwise, we perform
augmentation until we get 500 images in total. While performing augmentation we save the
augmented images to the same folder with the name
[originalname]_augmented_[count].[extension].

You might also like