Final Work Submission | GSoC’19 | CCExtractor Development
Hello folks,
And thus, GSoC ends. So, I guess this marks one of the last blog posts for the Google Summer of Code 2019. And, now the task is to finally submit my work. With this blog post, I’d like to comprehensively document all of my work done through the entire Google Summer of Code 2019. This post contains all the links to major PRs and repository.
Repositories:
Poor Man’s Rekognition:
GitHub Link: https://github.com/pymit/Rekognition
Front-end for Poor Man’s Rekognition:
GitHub Link: https://github.com/pymit/RekoUI
- Pull Requests: in Poor Man’s Rekognition
Merged PRs
- #4: addition of FaceNet lib
- #5: Multiface Detection & Recognition
- #6: Integration of DjangoRestFramework for creating RESTful APIs for image and shifting of video processing from cv2 with scikit-video.
- #7: RESTful APIs for video enhancement in video processing
- #8: Refactoring the codebase : Code Refactor
- #10: Video feature enhancement and code optimisation + #11 (related)
- #13: RESTful APIs for Embeddings (GET & POST methods)
- #14: fixing corsheader issue
- #15: Django models and serializers
- #16: improvement of Django models
- #22: Integration of redis and celery & asynchronous execution
- #24: Adding Docstrings & related PRs #34 #36 #38 #43
- #31: creating Feedback feature’s RESTful APIs #37: related PR
- #33: Facial Expression integration
- #39: face Recognition in YouTube videos
- #40: NSFW Classifier Integration
- #41: Face Comparision Feature
Unmerged PRs:
- #45: addition of unit test [Work in progress]
- #44: Django user authentication [Work in progress]
2. Pull Requests: in Front-end for Poor Man’s Rekognition:
Merged PRs
- #1: Listing all available embeddings or faceID, create and show using cards, and integration of Image & Video View components.
- #3: UI for embedding creation
- #4: Feedback feature + Routing + Basic UI for video playback & related PR#6
- #8: Similar Face
- #9: NSFW component in main Image view component.
- #10: API Documentation and view for How to use it. Related PR #13
Unmerged PRs:
#14: User Authentication [WIP]
All commits https://github.com/pymit/Rekognition/commits/master
How to setup this project locally?
This wiki link will guide you set up the project locally.
Links to previous report blogs
- Blog Link 1: GSoC with CCExtractor Development
- Blog Link 2: Project Architecture | Tools Selection
- Blog Link 3: Video Processing | Embedding Visualization
- Blog Link 4: Video Processing Improvement
- Blog Link 5: Frontend with ReactJS
- Blog Link 6: Celery | Redis | Asynchronous Execution
- Blog Link 7: TensorFlow Serving | Facial Expression
- Blog Link 8: Final Phase work submission
- Blog Link 9: Final Report
Future with Rekognition
These past three months were a journey of a steep learning curve. A lot of things has been achieved in this period but yet the path doesn’t end here. Lots of improvement and the integration of new features can be achieved in this project. Here are few things which I think will improve this project.
- A more robust video processing method.
- Integration of object scene detection.
- Natural language processing
- Docker & Kube support for the main repository.
- Better unit tests.
- Deployment of the project
GitHub kanban board link which lists all the future upgrades, work in progress and all the works done with corresponding commits and PRs mentioned.
Conclusion
I’ve had a wonderful time during these 3 months and have learned plenty of things. I am thankful to my primary mentor Johannes Lochter, and Carlos Fernandez Sanz , for providing me this opportunity and all the help and support they provided. I’m also grateful to all the members of CCExtractor Development for their quick responses, open interactions, healthy and constructive discussions.
And most importantly, I express my gratitude to the people behind the Google Summer of Code. I had a very enjoyable experience working in GSoC 2019 with CCExtractor Development. I will definitely keep contributing in open-source world!
Happy Coding!