week 1 | GSoC’19 | CCExtractor Development
Hello guys, I’m here with my second blog on reaching the first-week milestone of the GSoC period. Firmly gripping the bow and holding the arrow in the right direction will certainly increase the probability of getting the arrow at the right aim.
I’m using this analogy to create a fine relation between project, the right tech tools, and project architecture design respectively which will help to accomplish the main goal of the project at the end.
My first week of the GSoC period was a great milestone for me not because I only baked some code but also took my time to give a glance at various current tech stacks available which can make the project more efficient. This exploring of tech stack was the main part of my GSoC Community Bounding period. Of course, There is no doubt that CCExtractor Development is a great community. This time they have taken a great initiative of distributing any technical book on demand to the selected students. What could be more blissful way better than this to start summer coding 😀. Wait, did I forget to introduce you to my project? err…😬 .
Okay, Let’s first start with the introduction of the project Poor Man’s Rekognition. This project aims at providing a free alternative to Amazon’s Rekognition services. The main aim is to provide simple and easy to use API that can quickly analyze any image. This time this project has been particularly focused on Face recognition. But I have planned to extend it furthermore ,I’ll will update this in coming weeks. You can also find a short description of the project at the official CCExtractor’s Website link. To discuss it furthermore you can join the slack channel of CCExtractor Development by requesting an invitation using link. Our mentor for this project is Johannes Von Lochter. He is well known for his research paper publications in International Conference on Machine Learning and Applications (ICMLA) and other great conferences as well :).
Technical insight: This week so far, I have done setting up the project locally and GitHub repo link . I’m using GitHub’s Kanboard to organize and manage the milestones of the project (link). Here’s a very basic sketch of the project structure,
It has lots of integration coming in future weeks (: excited) and will update it on coming weeks or you can also keep eye on the kanban board. After setting up the environment, I also integrated some of the already implemented features of proof of concept of the PMR (Poor Man’s Rekognition). This week I worked on extending the support for png image format so that tensor input could easily fit the network architecture. I also worked on multi-face detection and recognition in a single frame merged PR#5 (Pull Request ) link . Currently, the project has a long way to be worked upon.
So, To further push the limits of enthusiasm for the project. I have planned to integrate the ReactJS for frontend. Week2 schedule: I will be focused on designing and creating the end RESTful APIs for the current features implemented using django-rest-framework. Concurrently, I’m also working on the video feature part and will update work around this in the coming blogs.
If you find this project intriguing then you can also start contributing. Stay tuned!
