Computer Vision - Object Detection on Videos - Deep Learning

Computer Vision - Object Detection on Videos - Deep Learning

This course provides an end-to-end coverage of Machine Learning on videos through Video analytics, Object Detection and Image Classification. It is a complete hands-on tutorial that teaches how to implement Video Analytics using the 3-step process of Capture, Process and Save Video, understand various Object Detection Models and implement them for a real-time case study of Social Distancing and last but not the least, take a deep dive into steps involved in using Deep Learning Models, Transfer Learning and learn how to create a model on face mask detection using Image Classification and leverage it to implement a solution of face mask detection.

This course is a must-have for all the developers in machine learning domain because of:

• Dedicated In-Course Support is provided within 24 hours for any issues faced

• Line-By-Line Code Walkthrough for object detection implementation on videos and training a model for image classification

• Comprehensive Coverage of Object Detection and Image Classification Models

• Working source code for People Footfall Tracking and Automatic Parking Management project

  
  

Below is the list of  key topics and projects we will be learning:

Video Analytics Architecture

In our course we are explaining in detail the complete architecture on which Video analytics operation are performed. It also covers in detail the functions available to perform various operations in Video like capturing video frame by frame and then saving these frames post operation into a running Video.

We also cover different supported Video Codec in which we can save videos post performing Machine Learning tasks.

Euclidean Distance

The course covers the implementation of Euclidean Distance to calculate the distance between different moving Objects, which in our case are humans.

Object Detection

We then move on to explain the different Object Detection Models available in the market starting from very basic model to understand the concept better which includes Haar Cascade and Hog Object Detection model, these two models uses simple mathematical and OpenCV code to locate object in the image. Post this, we move on to Deep Learning based Object detection models like RCNN and FAST RCNN. Then we move on to explain the logic behind Faster RCNN and R-FCN models. The more popular models SSD and Yolo V3 are also explained in detail to provide complete coverage of Object Detection

In our course we are also providing step by step code walk through for Haar Cascade and Hog implementation for detecting Human beings which are recognized as Objects in the video. Complete running code is provided for download along with this course. Now in the next step, we are also providing complete code walkthrough for identification of human beings in the same Video as used above but this time using Yolo V3 Deep Learning Model. Again, same Human detection model along with distance calculation is also provided using Faster RCNN model. Complete running code is available for download for YoloV3 and Faster RCNN with this course.

We are also explaining the concept of Transfer Learning which is integral part of all Object Detection and Image Classification solution for better accuracy.

Image Classification

To continue enhancement and provide robust coverage we have are also providing details on Image Classification models VGG16 and VGG19 and then we continue to explain more advanced models like ResNet50 and ResNet10. Later this journey is continued with InceptionV3 and Xception which are Deep Learning based models

Next, we are covering on how to train our own Models using our own dataset on Google Colab using Transfer Learning with the help of Tensorflow and Keras in which we are using InceptionV3 for Image Classification and classify them into two different class one with mask and one without mask. This trained model in then saved and used along with another Object Detection model to develop complete working solution.

Combining Both Object Detection and Image classification

In most of the solutions, Object Detection and Image classification goes hand in hand and in our example given in this course we are showing that with live demo.

Here, an input recorded video is provided within which we are viewing the people walking around. As shown during code walkthrough, we first detect faces of all person in the video with the help of Object Detection model and then using our own trained model we are performing image classification operation to finally detect if the person in the video is wearing mask or not. In this example we also perform different marking on different faces in video based on their classification.

We are also showing how by tuning few parameters we can perform similar task as mentioned above of identification and marking on LIVE Feeds as well.

Object Tracking

We are also teaching how we can perform object tracking operation on Videos with the help of Object Detection and Tracking algorithms. We are explaining this concept with the help of code walkthrough wherein we are making use of SORT framework to track object in consecutive frames and then with the help of marking on frames we can show the movement of object throughout the video.

People Footfall Tracking and Automatic Parking Management project is used as an example for showcasing Object tracking on recorded vide, the same concept can be implemented on Live feed as well.

Quick Starter on Object Detection and Image Classification on Videos using Deep Learning, OpenCV, YOLO and CNN Models

Url: View Details

What you will learn

• Learn how to implement Video Analytics using Deep Learning
• Understand how to implement Object Detection Models on Videos using Python
• Build your own Deep Learning model using Transfer Learning for Image Classification

Rating: 4.5

Level: All Levels

Duration: 2.5 hours

Instructor: Vineeta Vashistha