add code for generating manual score

This commit is contained in:
lukas-heiligenbrunner 2022-07-05 19:26:00 +02:00
parent c56e583f68
commit 1add9d278f
3 changed files with 37 additions and 5 deletions

View File

@ -1,11 +1,14 @@
import pickle
import numpy as np
import torch
from PIL import Image
import Compress
import DataLoader
import ex4
from ImageImpaint import get_train_device
from netio import load_model
from netio import load_model, eval_evalset, write_to_pickle
def apply_model(filepath: str):
@ -26,6 +29,37 @@ def apply_model(filepath: str):
im = Image.fromarray(out)
im.save("filename.jpg", format="jpeg")
def test():
# read the provided testing pickle file
print("Generating pickle file with privided test data")
PICKEL_PATH = "test"
model = load_model()
model.eval()
loader,_ = DataLoader.get_image_loader("training/", np.float32)
outarr = np.zeros(dtype=np.uint8, shape=(8663, 3, 100, 100))
targetarr = np.zeros(dtype=np.uint8, shape=(8663, 3, 100, 100))
i = 0
for input, target in loader:
out = model(input)
out = DataLoader.postprocess(out.cpu().detach().numpy())
outarr[i] = out
targetarr[i] = DataLoader.postprocess(target.cpu().detach().numpy())
print(f'\rApplying model [{i}/{len(loader)}]', end='')
i += 1
if i==8663:
break
write_to_pickle(PICKEL_PATH + "_pred.pkl", list(outarr))
# compress the generated pickle arr
Compress.compress(PICKEL_PATH + "_pred.pkl")
write_to_pickle(PICKEL_PATH + "_target.pkl", list(targetarr))
# compress the generated pickle arr
Compress.compress(PICKEL_PATH + "_target.pkl")
if __name__ == '__main__':
apply_model("training/000/000017.jpg")
eval_evalset()
# test()

View File

@ -115,8 +115,8 @@ def plot(inputs, targets, predictions, path, update, epoch):
for ax, data, title in zip(axes, [inputs, targets, predictions], ["Input", "Target", "Prediction"]):
ax.clear()
ax.set_title(title)
# ax.imshow(DataLoader.postprocess(np.transpose(data[i], (1, 2, 0))), interpolation="none")
ax.imshow(np.transpose((data[i]), (1, 2, 0)), interpolation="none")
ax.imshow(DataLoader.postprocess(np.transpose(data[i], (1, 2, 0))), interpolation="none")
# ax.imshow(np.transpose((data[i]), (1, 2, 0)), interpolation="none")
ax.set_axis_off()
fig.savefig(os.path.join(path, f"{epoch:02d}_{update:07d}_{i:02d}.png"), dpi=100)

View File

@ -27,8 +27,6 @@ import zipfile
import dill as pkl
import numpy as np
import onnx
import onnxruntime
TEST_DATA_PATH = r"/daten/challenge/django/data/datasets/image_inpainting_2022/test.zip"