Minimal_Example.py

import os
import subprocess
import time
import sys


def make_call_string(arglist):
    result_string = ""
    for arg in arglist:
        result_string += "".join(["--", arg[0], " ", arg[1], " "])
    return result_string


root_folder = os.path.dirname(os.path.abspath(__file__))
data_folder = os.path.join(root_folder, "Data")
model_folder = os.path.join(data_folder, "Model_Weights")
test_folder = os.path.join(data_folder, "Minimal_Example")
cropped_folder = os.path.join(test_folder, "Cropped")
street_view_folder = os.path.join(test_folder, "Street_View")
if not os.path.exists(cropped_folder):
    os.mkdir(cropped_folder)

# First download the pre-trained weights
download_script = os.path.join(model_folder, "Download_Weights.py")

print("Downloading Pretrained Weights")
start = time.time()
call_string = " ".join(
    [
        "python",
        download_script,
        "1aPCwYXFAOmklmNMLMh81Yduw5UrbHqkN",
        os.path.join(model_folder, "Houses", "trained_weights_final.h5"),
    ]
)

subprocess.call(call_string, shell=True)

call_string = " ".join(
    [
        "python",
        download_script,
        "1FbvHzQWCjucXPbTbI4S1MnBLkAi58Mxv",
        os.path.join(model_folder, "Openings", "trained_weights_final.h5"),
    ]
)

subprocess.call(call_string, shell=True)
end = time.time()
print("Downloaded Pretrained Weights in {0:.1f} seconds".format(end - start))

# Now run the housing detector
detector_script = os.path.join(
    os.path.dirname(
        os.path.abspath(__file__)),
    "2_Computer_Vision",
    "detector.py")


houses_result_file = os.path.join(test_folder, "Housing_Results.csv")
houses_weights = os.path.join(
    model_folder,
    "Houses",
    "trained_weights_final.h5")
houses_classes = os.path.join(model_folder, "Houses", "data_classes.txt")
anchors = os.path.join(
    root_folder,
    "2_Computer_Vision",
    "src",
    "keras_yolo3",
    "model_data",
    "yolo_anchors.txt",
)

arglist = [
    ["input_images", street_view_folder],
    ["classes", houses_classes],
    ["output", test_folder],
    ["yolo_model", houses_weights],
    ["box_file", houses_result_file],
    ["anchors", anchors],
]
call_string = " ".join(["python", detector_script, make_call_string(arglist)])

print("Detecting Houses by calling \n\n", call_string, "\n")
start = time.time()
subprocess.call(call_string, shell=True)
end = time.time()
print("Detected Houses in {0:.1f} seconds".format(end - start))

# #Next, we crop out the houses
cropping_script = os.path.join(
    root_folder,
    "2_Computer_Vision",
    "Crop_Images.py")


cropping_result_file = os.path.join(test_folder, "Cropping_Results.csv")


arglist = [
    ["input_file", houses_result_file],
    ["classes", houses_classes],
    ["output_folder", cropped_folder],
    ["output_file", cropping_result_file],
]


call_string = " ".join(["python", cropping_script, make_call_string(arglist)])

print("Cropping Houses by calling \n\n", call_string, "\n")
start = time.time()
subprocess.call(call_string, shell=True)
end = time.time()
print("Cropped Houses in {0:.1f} seconds".format(end - start))


# Next run the opening detector

model_folder = os.path.join(data_folder, "Model_Weights")
opening_weights = os.path.join(
    model_folder,
    "Openings",
    "trained_weights_final.h5")
opening_classes = os.path.join(
    model_folder,
    "Openings",
    "data_all_classes.txt")
detector_script = os.path.join(root_folder, "2_Computer_Vision", "detector.py")
openings_result_file = os.path.join(test_folder, "Opening_Results.csv")

arglist = [
    ["input_images", cropped_folder],
    ["classes", opening_classes],
    ["output", test_folder],
    ["yolo_model", opening_weights],
    ["box_file", openings_result_file],
    ["anchors", anchors],
    ["postfix", "_opening"],
]
call_string = " ".join(["python", detector_script, make_call_string(arglist)])

print("Detecting Openings by calling \n\n", call_string, "\n")
start = time.time()
subprocess.call(call_string, shell=True)
end = time.time()
print("Detected Openings in {0:.1f} seconds".format(end - start))

# Finally run the classification

classifier_script = os.path.join(
    root_folder,
    "3_Classification",
    "Classifier.py")

level_folder = os.path.join(data_folder, "Level_Detection_Results")
softness_score_file = os.path.join(test_folder, "Softness_Scores.csv")

arglist = [
    ["output_file", softness_score_file],
    ["input_file", openings_result_file],
    ["level_folder", test_folder],
    ["classes", opening_classes],
]

call_string = " ".join(
    ["python", classifier_script, make_call_string(arglist)])


print("Calculating Softness Scores by calling \n\n", call_string, "\n")
start = time.time()
subprocess.call(call_string, shell=True)
end = time.time()
print("Calculated Softness Scores in {0:.1f} seconds".format(end - start))