VEHICLE COLLISION AVOIDANCE UNDER REDUCED VISIBILITY CONDITIONS PRATEEK
VEHICLE COLLISION AVOIDANCE UNDER REDUCED VISIBILITY CONDITIONS PRATEEK AGGARWAL
OUTLINE Discussion about previous work Framework used YOLO Why YOLO? How it works ? Applications of YOLO My project work Work samples 2
PREVIOUS WORK Rear End Collisions – A rear end collision occurs when one car hits the car in front of it from behind Impacts of Rear End Collisions and losses related to it. Reasons of Rear End Collisions – common reasons like driver inattention or distraction , tailgating, panic stops , and reduced visibility due to weather or worn pavement. Prerequisites
OUTLINE Discussion about previous presentation Framework used YOLO Why YOLO? How it works ? Applications of YOLO My project work Work samples 4
FRAMEWORK USED Darknet 1. It is an open source neural network framework. 2. It is fast and supports CPU and GPU computation. 3. It can be used for many thing like YOLO , Image. Net Classification , RNNs , training a classifier on CIFAR – 10.
OUTLINE Discussion about previous presentation Framework used YOLO Why YOLO? How it works ? Applications of YOLO My project work Work samples 6
YOLO You Only Look Once It is a state-of-the-art , real-time object detection system. It is a system for detecting objects on the Pascal VOC 2012 dataset. Prior work on object detection repurposes classifiers to perform detection. YOLO model processes images in real-time at 45 frames per second to 155 frames per second.
YOLO The Unified architecture of YOLO model is extremely fast than other models. The base YOLO model can process images in real time at 45 frames per second. The fast YOLO model , which is a smaller version of the network can process with an astounding speed of 155 frames per second without losing the quality of image processing.
OUTLINE Discussion about previous presentation Framework used YOLO Why YOLO? How it works ? Applications of YOLO My project work Work samples 11
WHY YOLO? Compared to other state-of-the-art object detection systems , YOLO makes more localization errors but it is far less likely to predict false detections where no object exists. YOLO learns very general representation of objects. It performs better than all other object detection methods , including DPM and R-CNN , by a wide margin when generalizing. Reference – https: //arxiv. org/abs/1506. 02640
WHY YOLO? The human visual system is fast and accurate, allowing us to perform complex tasks like driving with little conscious thought , but sometimes that could be dangerous. Fast, accurate algorithms for object detection would allow computers to drive cars without specialized sensors, enable assistive devices to convey real-time scene information to human users, and unlock the potential for general purpose, responsive robotic systems.
WHY YOLO? Moreover, we can train it from scratch according to our needs for detection of any kind of object that is present in VOC dataset from 2007 to 2012. Combining with CUDA and open. CV we can use It with web cam to perform real-time object detection in real life scenarios. Reference – https: //github. com/pjreddie/darknet/wiki/YOLO: -Real-Time-Object-Detection
OUTLINE Discussion about previous presentation Framework used YOLO Why YOLO? How it works ? Applications of YOLO My project work Work samples 15
HOW IT WORKS ? It uses a totally different approach than other detection systems. It applies a single neural network to the full image. The network divides the image into regions and predicts bounding boxes and probabilities for each region the bounding boxes are weighted by the predicted probabilities. If the threshold value or the confidence ranging from 0 to 1 is higher than. 25 it then performs detection otherwise it does not. We can change this value according to our necasity.
OUTLINE Discussion about previous presentation Framework used YOLO Why YOLO? How it works ? Applications of YOLO My project work Work samples 18
APPLICATIONS OF YOLO Unified detection. Network design. Training. Experimentation.
OUTLINE Discussion about previous presentation Framework used YOLO Why YOLO? How it works ? Applications of YOLO My project work Work samples 20
MY PROJECT WORK I have used YOLO and darknet to detect vehicles like cars and trucks under various circumstances of weather like fog , rain , storm , etc. It worked fine and fast on CPU. In approximately 20 seconds using 30 to 31 layers it predicted vehicles.
OUTLINE Discussion about previous presentation Framework used YOLO Why YOLO? How it works ? Applications of YOLO My project work Work samples 22
THANK YOU. .
- Slides: 31