XiaopinWang/mask_r_cnn_pig_detection

Pig detection project - Count pig's numbers and Seize their behavior pattern(standing,lying).

Ultimately We will get pig's weight(area of pig's fixel??) with image with masking.

0.Use MatterPort's Mask R-CNN Code

1.We start from balloon.py (1 class)

2.We need 2 classes (standing pig,lying pig)

3.So we refer to https://github.com/SUYEgit/Surgery-Robot-Detection-Segmentation

4.I seperate train,test code.(train.py ,test.py) Because we need understand code.

5.Now we combine these code again(main.py), and make GUI with PyQt5

We change mrcnn/visualize.py -> display_instances method ( save img file here ) In pig_images/ json file "region_attributes":{"name" : "standing_pig"} <---- class name We use VGG Image Annotation tool http://www.robots.ox.ac.uk/~vgg/software/via/via-1.0.6.html

Not complete project (19/09/18)