To detect any reasonable change in a live cctv to avoid large storage of data. Once, we notice a change, our goal would be track that object or person causing it. We would be using Computer vision concepts. Our major focus will be on Deep Learning and will try to add as many features in the process.
DAY 1: Implemented the cosine difference algorithm in video to calculate similarity between 2 consecutive frames of a video
DAY 2: Implemented euclidean distance and manhattan diatance to realtime video. The program was also set to record the video
if significant changes were detected in any frame
DAY 3: Tried to implement the YOLO algorithm with pretrained models for object detection
DAY 4: Implemented cv2 functions to detect object in an image and make a bounding box around it.
DAY 5: The code now detectes moving object in a realtime and also records if significant changes are detected in the frames of
the video
DAY 1(13/09/21): Finished Linear algebra(16/16) and neural network(3/3) playlist (3B1B)
Held a meet with seniors regarding the project, ideas and the way to
move forward.
DAY 2(14/09/21): Saw some research papers on scene change detection. Will look for more
and finalize some to read.
DAY 3(15/09/21): Started with the Coursera course on Deep learning (first week).
DAY 4(16/09/21): Completed first week of the 1st course, started with the second week.
DAY 5(17/09/21): Completed the second week of the course and also downloaded VM ware and ubuntu
20.04 LTS and set it up. Had a meet with the mentor regarding the way
ahead with the project.
DAY 1: Coursera course 4 started, also reading reasearch paper regarding the project.
DAY 2: Looked about YOLO and many more algorithms for object detection. Continued with the coursera course.
Learned more about openCV from various sources
DAY 3: Coursera course 4 week 1 completed with ideas of basic CNN.
DAY 4: Learned about Cosine similarity and implemented it to calculate similarity between 2 images. Started to study the YOLO
algorithm.
DAY 5: Studied YOLO algorithm and also read and implemented euclidean and manhattam distance for image similarity.