History and architecture of YOLO v5

Yolo V5 Architecture

CNN-based Object Detectors are primarily applicable for recommendation systems. YOLO (You Only Look Once) models are used for Object detection with high performance. YOLO divides an image into a grid system, and each grid detects objects within itself. They can be used for real-time object detection based on the data streams. They require very few computational resources.

History of YOLO


Simple code to display grayscale thermal images

Photo by USGS on Unsplash

Thermal images are used in different fusion techniques for Object detection in Autonomous Vehicles (AVs). These images are in grayscale. We can use simple code in Python to view these images for analysis. There are some publicly available datasets that have Thermal images for machine learning in AVs.

# open cv
import cv2
# plotting library
import
matplotlib.pyplot as plt
image = cv2.imread('flir_thermal.jpeg', 0)
colormap_image = cv2.applyColorMap(image, cv2.COLORMAP_TWILIGHT_SHIFTED)

plt.figure()
plt.imshow(colormap_image)
plt.show()

CV2 has the following Color Maps:

COLORMAP_AUTUMN
COLORMAP_BONE
COLORMAP_JET
COLORMAP_WINTER
COLORMAP_RAINBOW
COLORMAP_OCEAN
COLORMAP_SUMMER
COLORMAP_SPRING
COLORMAP_COOL
COLORMAP_HSV
COLORMAP_PINK
COLORMAP_HOT
COLORMAP_PARULA
COLORMAP_MAGMA
COLORMAP_INFERNO
COLORMAP_PLASMA
COLORMAP_VIRIDIS
COLORMAP_CIVIDIS
COLORMAP_TWILIGHT
COLORMAP_TWILIGHT_SHIFTED
COLORMAP_TURBO
COLORMAP_DEEPGREEN

The following sample images can be downloaded from here.


Photo by Brock Wegner on Unsplash

Several barriers hinder the widespread acceptance of Autonomous Vehicles (AVs). Consumer perception, social influence, system characteristics, geographic and economic factors influence adoption. Some of the key challenges are listed below:

1 - Human behavior: Consumer Trust, Decision making & Control

Humans trusting AVs to execute key decisions may take a longer period. AVs have been in the market for almost a decade with significant hands-free improvements. Yet, we have seen several real-life situations wherein drivers doze off or ignore ongoing traffic needs resulting in manufacturers adding in mandatory check-in every few minutes to ensure drivers are attentive. …


Steps to detect an object using a sample dataset

In general, the Classification technique does not help much in Autonomous Vehicles as it predicts only one object in an image and does not give the location of that image. Object detection is very important in Autonomous Vehicles to detect what objects are there in a scene and their location. YOLO (You Only Look Once) is a high-speed real-time Object Detection algorithm created by Joseph Redmon et al. as a regression problem providing class probabilities and uses convolutional neural networks (CNN). Later, it has undergone some revisions. To make predictions, YOLO…


Simple steps to create an automated folder structure!

Photo by Kevin Ku on Unsplash

Having a well-organized general Machine Learning project structure makes it easy to understand and make changes. Moreover, this structure can be the same for multiple projects, which avoids confusion. In this post, we will use the Cookiecutter package to create a Machine Learning project structure.

Step 1: Make sure that you have latest python and pip installed in your environment.

Step 2: Install cookiecutter

pip install cookiecutter

Step 3: Create a sample repository on github.com (e.g., my-test)


Car vector created by macrovector — www.freepik.com

AVs struggle to identify humans with dark skin tone¹ alongside the vehicle or walking in front of them. This could lead to situations where they fail to see police officers, pedestrians, or workers on the side of construction zones giving instructions. AVs have a tough time in bad weather conditions in which an entire array of problems are created for the algorithms to solve.

So far, there are no unified ethical standards and certifications² for AVs. The big Moral Machine study³ conducted by MIT showed that it’s hard to identify universal ethical values. The moral choices that people made in…


https://www.freepik.com/vectors/car

There are many benefits due to Autonomous Vehicles. Please go through the post ‘Benefits of Autonomous Vehicles.’ However, as AVs collect camera images, video clips, biometric data, user contacts, location, speed, date and time, owner or passenger information, navigation history, etc., there is a risk to privacy.

Owner and Passenger Information

AVs collect a large amount of data, including the owner & passenger information, audio and video content inside the cabin and its surroundings, biometric data, map data, personal preference information such as seat inclination, temperature settings, etc. This information can identify drivers, passengers, and their activities with a high degree of certainty.

Location data

The…


Datasets with multiple sensor modalities (LiDAR, RADAR, Stereo Camera, Thermal Camera etc.)

Car vector created by upklyak — www.freepik.com

A wide variety of sensors are used in autonomous vehicles. The diversity of sensing modalities helps in different weather conditions. The following is a popular list of autonomous driving datasets which have been published up to date.

Note: As new datasets are being released every couple of months, I will be updating this page every month if a popular dataset is released.


AV Recommendations

A variety of measures may be employed to help protect personal information collected and stored by Autonomous Vehicles (AVs). The following are the recommendations:

1) Privacy by design

The design of an AV can include privacy by design, right from the beginning of the product development process, considering the context and content. The data collected by an AV have the complexity arising from multiple drivers, rental vehicles, change of ownership, and others as they leave traces of their data in the vehicle.

The FTC endorsed privacy by design¹ and called for entities to implement best practices to protect consumers’ private information. …


Car vector created by freepik — www.freepik.com

Autonomous Vehicles (AVs) have been automating various aspects of driving, including adaptive cruise control, parking assistance, lane centering, driver assist, collision avoidance systems, automated brake assistance towards a continuous process improvement geared towards automotive safety.

AVs are rated according to the SAE’s levels of automation:

Surya Gutta

Software Architect | Machine Learning | Statistics | AWS | GCP

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store