Computer Vision Project - Detect Activities

Problem statement: Create a program which can detect:- (a) If a person is smoking. (b) If a person is spitting. It should be an online program, i.e. should be able to able to discern the state as a video is being fed to it. There may be multiple people in the video, the person's face should be highlighted and captured. For (a), the following activities can be thought to constitute smoking: 1. Having a lit cigarette. 2. Any action that indicates the lighting of a cigarette. 3. Actually smoking. 4. Exhaling smoke(no cigarette in sight) (b) is self-explanatory. The project should be cloud deployable(heroku/pythonanywhere) and a user should be able to give it videos via the following mechanism: 1. Via a UI 2. Via an API For any video given to the program, a report should be generated in the following format: A zipped file per person(contains the face + the timestamp at which the person has smoked/spit) Any technology/APIs can be used as long as the above conditions regarding the problem statement are met. The developer should be willing to sign an NDA and an IP agreement. When sending a bid, please describe the approach you would take to solve the project along with time bound milestones and deadlines. Also please describe your previous projects. Apart from (a) and (b), the ability to add any other classifier should be present(in a plug-and-play manner). Say that one wants to add a way to detect if two people are fighting and a classifier is already available, one should be able to do the same with minimal code changes.