def displaySpecificPredictData(self,
predictor,
path_to_page,
category=None):
print(path_to_page)
im = cv2.imread(path_to_page)
outputs = predictor(im)
v = CustomVisualizer(im[:, :, ::-1], scale=1)
v = v.draw_instance_predictions(outputs["instances"].to("cpu"),
category)
self.cv2_imshow(v.get_image()[:, :, ::-1])
def displayRandomSampleData(self,
data,
meta_data,
page_count,
category=None):
for d in random.sample(data, page_count):
print(d["file_name"])
img = cv2.imread(d["file_name"])
visualizer = CustomVisualizer(img[:, :, ::-1],
metadata=meta_data,
scale=1)
vis = visualizer.draw_dataset_dict(d, category)
self.cv2_imshow(vis.get_image()[:, :, ::-1])
def displaySpecificSampleData(self,
data,
meta_data,
path_to_page,
category=None):
d = [x for x in data if x["file_name"] == path_to_page][0]
print(d["file_name"])
img = cv2.imread(d["file_name"])
visualizer = CustomVisualizer(img[:, :, ::-1],
metadata=meta_data,
scale=1)
vis = visualizer.draw_dataset_dict(d, category)
self.cv2_imshow(vis.get_image()[:, :, ::-1])
def displayRandomPredictData(self,
predictor,
data,
meta_data,
sample_ammount=3,
category=None):
for d in random.sample(data, sample_ammount):
print(d["file_name"])
im = cv2.imread(d["file_name"])
outputs = predictor(im)
v = CustomVisualizer(im[:, :, ::-1], metadata=meta_data, scale=1)
v = v.draw_instance_predictions(outputs["instances"].to("cpu"),
category)
self.cv2_imshow(v.get_image()[:, :, ::-1])
def predict_image(predictor, img_file, display_original_image, display_multiple_classes, which_classes):
image = Image.open(img_file).convert("RGB")
if display_original_image:
st.image(image, "Your uploaded image", use_column_width=True)
im = np.array(image)
outputs = predictor(im)
if display_multiple_classes:
for c in which_classes:
v = CustomVisualizer(im[:, :, ::-1], scale=1)
v = v.draw_instance_predictions(outputs["instances"].to("cpu"), [c])
st.image(v.get_image()[:, :, ::-1], use_column_width=True)
v = CustomVisualizer(im[:, :, ::-1], scale=1)
v = v.draw_instance_predictions(outputs["instances"].to("cpu"))
st.image(v.get_image()[:, :, ::-1], use_column_width=True)
#%%
trainer = DefaultTrainer(cfg)
trainer.resume_or_load(resume=True)
predictor = DefaultPredictor(cfg)
# %%
# step by step process to calculate stave measures from the system measures and staves:
# fetch random raw prediction boxes:
for d in random.sample(muscima_data, 1):
im = cv2.imread(d["file_name"])
outputs = predictor(im)
print(d["file_name"])
print(outputs["instances"].pred_boxes)
v = CustomVisualizer(im[:, :, ::-1], metadata=metadata, scale=1)
v = v.draw_instance_predictions(outputs["instances"].to("cpu"))
ImageDisplayer().cv2_imshow(v.get_image()[:, :, ::-1])
# %%
# get all boxes for one image:
path_to_img = "./../Data\CVC_Muscima_Augmented\CVCMUSCIMA_MultiConditionAligned\staffline-y-variation-v1\w-25\p012.png"
ImageDisplayer().displaySpecificPredictData(predictor, path_to_img, [0])
ImageDisplayer().displaySpecificPredictData(predictor, path_to_img, [1])
im = cv2.imread(path_to_img)
outputs = predictor(im)
all_boxes = outputs["instances"].pred_boxes.tensor.cpu().numpy()
all_classes = outputs["instances"].pred_classes.cpu().numpy()
print(all_boxes)