Do users need a free, open source surveillance app power by state of the art computer vision(deep learning) technology?
tham last edited by
I am planning to develop an surveillance app like this one. Not sure app like this needed by users or not.
Main purposes are
- Make it part of my portfolio
- Sharpen my skills
Functions plan to support for 1.0
- Emit alarm if detect certain objects(people, car, dog etc)
- Collect data of person, like age and gender, show the data on chart
- Emit alarm if detect suspicious person in the database(by face recognition)
- Access webcam and cctv
- Support gpu/cpu
- Send alert info with image to email
- ROI(region of intereset) selection
Another interesting functions like smoke and garbage detection are considering. I think smoke detection of indoor is a doable with semantic segmentation, outdoor could be more complicated since I believe cloud and fog are harder to discriminate.
Garbage detection is another topic, I am not sure it is doable or not, you can check the discussion at mxnet forum if you are interested.
I can handle most of the technical issues by myself, including gui, computer vision algorithm, deep learning, network etc, but it wouldn't have beautiful ui since I am not good at art.