Python for Image and Video processing applications
Download as PPTX, PDF1 like322 views
The document describes a machine learning project to classify contaminated drinking water samples using images taken with a Raspberry Pi embedded system. A KNNMPCAF3 algorithm is used to extract features from the images and classify them using k-nearest neighbors. The system is integrated with a Blynk IoT app to allow remote monitoring and control via a mobile phone. Accuracy of over 76% was achieved in classifying four levels of contamination. The overall aim is to create a low-cost solution to help monitor water quality.
![Software Installation
Python 3.7
* Download Python latest version from Python.org
* Install the Python in windows in any drive
* After installation check whether it is properly installed or not
in your system using the command
D:Program FilesPython37>python
Python 3.7.0 (v3.7.0:1bf9cc5093, Jun 27 2018,
04:06:47) [MSC v.1914 32 bit (Inte
l)] on win32
Type "help", "copyright", "credits" or "license" for more
information.
>>>](https://image.slidesharecdn.com/ramachadracollegeofengineering-200707164854/85/Python-for-Image-and-Video-processing-applications-4-320.jpg)
![Software Installation
Numpy, Scipy and Matplotlib installation
D:Program FilesPython37>python
Python 3.7.0 (v3.7.0:1bf9cc5093, Jun 27 2018, 04:06:47) [MSC
v.1914 32 bit (Inte
l)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy
>>> import scipy
>>> import matplotlib
>>>
1. python -m pip install numpy
2. python -m pip install scipy
3. python -m pip install matplotlib](https://image.slidesharecdn.com/ramachadracollegeofengineering-200707164854/85/Python-for-Image-and-Video-processing-applications-5-320.jpg)
Python for Image and Video processing applications
- 1. Python for Image Processing Applications One Week Faculty Development Program on Research Challenges in Computer Science & Engineering – RCCSE-2020 01.06.2020 to 05.06.2020 RAMACHANDRA College of Engineering Eluru, Andra Pradesh Presented by, Dr.R.Senthilkumar, Assistant Professor, Dept. of ECE, Institute of Road and Transport Technology, Erode, Tamil Nadu-638316 Email: rsenthil.optical@gmail.com, Contact no: 9940882605 Tuesday, July 7, 2020
- 2. Python Libraries Python 3.7 and above Matplotlib Numpy Scipy Pillow OpenCV Scikit-learn
- 3. Software Weblinks Python 3.7 - https://www.python.org/ Matplotlib - https://matplotlib.org/ Numpy - https://numpy.org/ Scipy - https://www.scipy.org/ Pillow - https://pypi.org/project/Pillow/ OpenCV - https://pypi.org/project/opencv-python/ Scikit-learn - https://scikit-learn.org/ (
- 4. Software Installation Python 3.7 * Download Python latest version from Python.org * Install the Python in windows in any drive * After installation check whether it is properly installed or not in your system using the command D:Program FilesPython37>python Python 3.7.0 (v3.7.0:1bf9cc5093, Jun 27 2018, 04:06:47) [MSC v.1914 32 bit (Inte l)] on win32 Type "help", "copyright", "credits" or "license" for more information. >>>
- 5. Software Installation Numpy, Scipy and Matplotlib installation D:Program FilesPython37>python Python 3.7.0 (v3.7.0:1bf9cc5093, Jun 27 2018, 04:06:47) [MSC v.1914 32 bit (Inte l)] on win32 Type "help", "copyright", "credits" or "license" for more information. >>> import numpy >>> import scipy >>> import matplotlib >>> 1. python -m pip install numpy 2. python -m pip install scipy 3. python -m pip install matplotlib
- 6. Run: pip install opencv-python - if you need only main modules Run: pip install opencv-contrib-python - if you need both main and contrib modules
- 7. Basic Image Processing using matplotlib & pillow Python libraries Exercises: 1. Image read and display 2. Pseudo color Image 3. Pseudo color Image color bar 4. Image Resizing 5. Image Interpolation 6. RGB to Gray Image 7. Histogram Plot 8. Cropping a Portion of an Image 9. Shape of an Image and gray scale conversion 10.Image transform 11.Image Filtering 12.Image Details and Changing Image File Format
- 28. More practice- refer the youtube link https://www.youtube.com/watch?v=Me2OWBstBNg&t=21s
- 29. Image Processing using OpenCV Exercises: 1. Read a colour and display an image 2. Read a colour image and display the size of an image 3. Convert a colout image into Gray image 4. Vertical and Horizontal stack more than one image 5. Image transform (rotation) 6. Image resize
- 30. Output: Exercise 1 Image size (183, 275) Output: Exercise 2
- 34. Output: Exercise 6 Orignal Image Resized Image
- 35. More practice- Refer the youtube video link https://www.youtube.com/watch?v=CCkDS-fo-eQ
- 36. Video Processing using OpenCV Exercises: 1. Capture a colour video from web camera 2. Capture a colour video from web camera and covert into gray video 3. Capture a colour vido and get its frame width and height 4. Set the user specified frame width and height 5. Play a already recorded video 6. Capture a video using webcamera and flip that video
- 37. Exercises: 7. Converting Colour video to Gray video 8. Converting Colour video to Gray and Gray to Binary video 9. Video Blurring (Low pass filtering) 10. Video resize and interpolation followed video blurring 11. Edge detection 12. Video Masking 13. Histogram Equalization 14. Video image transform 15. Video motion Detection
- 38. More practice –Refer the youtube video lecture https://www.youtube.com/watch?v=bR01_iGx7os
- 39. Machine Learning based Classification of Contaminated Drinking Water Using Raspberry Pi Embedded System and IoT Device Aim of this research project: The drinking and human usage fresh water are contaminated and polluted by wastages released from dyeing and other industries. The classification of highly contaminated, mildly contaminated, lightly contaminated and recycled water is important for different level of water purification. The aim of this project is to develop an Internet of Things based low cost embedded Raspberry pi model for drinking water image classification. For image classification, a machine learning based algorithm proposed here called as KNNMPCAF3 (K-Nearest Neighbour Machine Learning based Principle Component Analysis First Three).
- 40. Algorithm KNNMPCA3 Algorithm Step 1: Start the program. Step 2: Collect the water sample images. Step 3: Derive top 3 principle components after pressing the keypad. Step 4: PCAs corresponding to each image samples are encoded. Step 5: The encoded image samples are classified using KNN classifier. Step 6:The classification accuracies are calculated for different number of neighbours. Step 7:Results are plotted.
- 41. Flowchart
- 42. Proposed Model Block Diagram
- 43. Collected River Water samples Non-Contaminated River Water Sample Lightly Contaminated River Water Sample Moderately Contaminated River Water Sample Severly Contaminated River Water Sample
- 45. Blynk IoT App configuration in Mobile Phone
- 46. Interface Raspberry pi Embedded kit Keypad, LEDs and Buzzer
- 47. Interfacing VNC viewer installed Laptop and Raspberry pi kit using USB
- 48. Principle Component Analysis Feature Extraction
- 49. Table 1: Images captured their respective classes and along with top 3 principle components are listed here Class Number and Class Name Image Id PCA 1 PCA 2 PCA 3 Class ‘0’ Non-contaminated water I1 0.639 0.609 0.594 I2 0.784 0.505 0.476 I3 0.639 0.609 0.594 Class ‘1’ Lightly contaminated water I4 0.874 0.871 0.836 I5 0.877 0.762 0.722 I6 0.167 0.144 0.141 Class ‘2’ Moderately contaminated water I7 0.133 0.133 0.129 I8 0.285 0.285 0.261 I9 0.134 0.132 0.129 Class ‘3’ Severely contaminated water I10 1.048 1.007 9.547 I11 0.487 0.461 0.418 I12 0.134 0.132 0.122 I13 1.611 1.500 1.247
- 50. Table 2: Raspberry pi general purpose input and output pins configuration and corresponding interfacing devices GPIO Pin Number Configured Input/output Connected Device GPIO 8 Input Keypad GPIO 10 Output BLUE Colour LED GPIO 16 Output GREEN Colour LED GPIO 18 Output RED Colour LED GPIO 22 Output BUZZER GPIO 6 Ground GND pin 3
- 51. Manual Control & Operation of Mobile Phont IoT App by a health inspector or sanitary worker
- 52. Software display- Python programmed and display the results in terminal or python shell script
- 53. Table 3: Recognition accuracy and false alarm for all the four cases Number of KNNs Recognition Accuracy in Percentage False Alarm in Percentage 1 100 0 2 76.92 23.08 3 76.92 23.08 4 69.23 30.77
- 54. Plot between No.of Nearest neighbor vs % Recognition Accuracy & False alarm
- 55. More details- refer YouTube video lecture 1. Machine learning using Python+Raspberry pi - https://www.youtube.com/watch?v=5SrZTyuFpmU&t=13s 2. Image classification using Python+Raspberry pi+Machine learning part 1 – https://www.youtube.com/watch?v=GpJomlEiJfQ&t=18s 3. Image classification using Python+Raspberry pi+Machine learning part 2– https://www.youtube.com/watch?v=jvLjRh2bGAI&t=51s 4. Image classification using Python+Raspberry pi+Machine learning part 3 – https://www.youtube.com/watch?v=YIjGiEWF7IY&t=15s 5. Machine Learning based Classification of Contaminated Drinking Water Using Raspberry Pi Embedded System and IoT Device https://www.youtube.com/watch?v=-oONEWCjLTw&t=369s
- 56. Is it necessary to file copyright or patent our work http://copyright.gov.in/frmStatusGenUser.aspx Diary number:6588/2020-CO/L