predictionTest

from deepforest import main
from deepforest import get_data
import os
import matplotlib.pyplot as plt
import numpy as np

model = main.deepforest()
model.use_release()

#imgPath = "C:\Hackathon Stadtbaeume\Data\DOP_20_C_ADV_25832_Quad_S.tif"
imgPath = "C:\Hackathon Stadtbaeume\Data\TestEdeka3000x3000.jpg"

# Show image or use Boxes output?
showIm = 

#Filter parameters
threshold_NMS = 0.1
sigma_NMS = 0.01
patchSize = 400
patchOverlap = 0.3


#threshold_NMSes = [0.1]
#sigma_NMSes = np.linspace(0,0.06,7)


#for threshold_NMS in threshold_NMSes:
if showIm:
    retPlot = True
else:
    retPlot = False

imOrBox = model.predict_tile(raster_path = imgPath,
                                        patch_size = patchSize, #default 500
                                        patch_overlap = patchOverlap, #default 0.3
                                        use_soft_nms=True,
                                        sigma=sigma_NMS,
                                        thresh=threshold_NMS,
                                        return_plot = retPlot)
if showIm:
    plt.imshow(imOrBox) #[:,:,::-1]
    #plt.show()
    
    #Save figure with filter parameters
    fname = r"C:\Hackathon Stadtbaeume\Testfigures\\" + "Edeka3000x3000" + "_threshNMS" + str(threshold_NMS) + "_sigmaNMS" + str(sigma_NMS) + "_patchSize"+ str(patchSize) + "_patchOverlap" + str(patchOverlap)+ ".png"
    plt.savefig(fname,dpi=80)
    #plt.savefig(fname,dpi=167)

else:
    print(imOrBox)