-
Notifications
You must be signed in to change notification settings - Fork 0
CHAPTER 6 : Technical Specifications
Numpy
- In Python we have lists that serve the purpose of arrays, but they are slow to process.
- NumPy aims to make array objects 50 times faster than regular Python lists.
- NumPy arrays are stored at one continuous place in memory unlike lists, so processes can access and manipulate them very efficiently.This is known as locality of reference in computer science.
Pandas
- Pandas is a widely used open-source Python tool for data science, data analysis, and machine learning.
- It is built on top of Numpy and provides support for multi-dimensional arrays.
- Pandas, as one of the most widely used data-wrangling tools, is compatible with a wide range of other Python data science modules.
OpenCV
- OpenCV is a comprehensive open-source library for computer vision, machine learning, and image processing.
- OpenCV can process images and videos to identify objects, faces, or even the handwriting of a human.
- OpenCV is a completely free programme.
- In our case, for the detection of pathoholes and for detection of objects like cars, motorcycles, peoples to analyze traffic, OpenCV library plays a very important role.
CV2
- OpenCV has a function to read video, which is cv2.
- cv2. The imread() method opens a file and loads an image from it.
- This function produces an empty matrix if the picture can't be read (due to a missing file, insufficient permissions, or an unsupported or incorrect format).
We are collecting raw data from two sources. The first one is using the OBD port of the car to collect vehicle acceleration data and the second is a smartphone which is used to collect the video data for image processing.
The data that is collected for the OBD port is saved in csv format in various categories for cleaning and preprocessing. The data that is collected from smartphones is stored locally and is separated frame by frame in order to pass it for image processing and training.There are errors, omissions, and discrepancies in the raw data. It necessitates modifications following a thorough examination of the completed questionnaires.
Data Preprocessing could be a technique that's accustomed to convert the {raw knowledge | data | information} into a clean data set. To put it another way, anytime data is received from several sources, it is acquired in a raw format that makes analysis impossible. We are utilizing mobile sensor information along with OBD data to build ML and analytics model. For the analytics part, data is collected from the OBD module from which we will select the suitable parameters for our applications. The data that has been filtered is passed to the modules to train them.
After passing the data to our modules the resulting numbers will be displayed on the smartphone.
The algorithms that we use to process the data collected from the ports and smartphone gives us accurate location of path-holes, the location of traffic, it shows us the severity of turns and what speed to maintain for safe turning, and it gives us the characteristics of the driver for better vehicle efficiency.
- The model used in the project will be trained as it will be used more in future. The more it gets trained, the better results we may obtain, so as to be more accurate.
- This feature can be used as a regular side gear in vehicles in future so as to avoid heavy traffic but primarily potholes.
- This may give users correct indications about the pothole or similar kind of anomalies on road; the user can slow down or drive carefully according to the instructions received from the model.
- The research based project can be used further for increased accuracy.