Concept
The Object Detection will be done through a webcam, generating live video.
The user will be able to choose between differrent models for detection, depending on the available.
This is important because different models require vastly more performance, while also delivering greatly increased detection-quality.
Currently being implemented are models of YOLO, and SSD
https://pjreddie.com/darknet/yolo/https://pjreddie.com/darknet/yolo/
https://github.com/weiliu89/caffe/tree/ssd
Performance-Comparison between popular models
For the GPS-data of the detected objects, I will only use a semi-random silumation to get data.
This is something that could be made with a real GPS-module in a later project, where this program gets onto mobile hardware
The user will, through the UI, be able to display the detected objects on a map and save the image.
The style will be similar to this:
The user can export the gatherd data as a file. A possible file type could be .csv
Concept may change in the future ;)