These are the dependencies: Tensorflow-gpu==1.15 opencv==3.* PIL numpy>=1.16.0
The project is based on the open-sourced tensorflow model for object detection which can be found in this link https://github.com/tensorflow/models/tree/master/research/object_detection
The greenpet.rar has the files for the Application on androidstudio.
Here is are examples on how our model generates stickers for flowers:
It is recommended to try this on a conda environment.
custom_image_detector.py is where you can put an image path and it will detect all the trees, plants, flowers there. custom_image_detector_webcam.py is where you provide your video link and it will iterate through different frames and do the same steps in custom_image_detector.py vis_stickers.py is where the real editing on the image happens. It was taken from the open-sourced tensorflow utils and made many changes in order to have stickers not rectangles surrounding the plant.