ffe

automation/assets_manager.py

@@ -137,7 +137,7 @@ def merge(self, fd: Path, td: Path):
 
 
 def run():
-    sources = "https://github.com/QIN2DIM/hcaptcha-challenger/issues/872"
+    sources = "https://github.com/QIN2DIM/hcaptcha-challenger/issues/826"
     am = AssetsManager.from_sources(sources)
     am.execute()
 

automation/auto_labeling.py

@@ -137,7 +137,7 @@ def run():
 
     for card in flow_card:
         # Filter out the task cards we care about
-        if "sedan_car" not in card["joined_dirs"]:
+        if "nested_smallest_turtle" not in card["joined_dirs"]:
             continue
         # Generating a dataclass from serialized data
         dl = DataLake(

automation/check_yolo_seg_model.py

@@ -0,0 +1,67 @@
+import os
+import shutil
+import sys
+from pathlib import Path
+
+import cv2
+import hcaptcha_challenger as solver
+from hcaptcha_challenger.onnx.modelhub import ModelHub
+from hcaptcha_challenger.onnx.yolo import YOLOv8Seg
+from tqdm import tqdm
+
+solver.install(upgrade=True)
+
+# Initialize model index
+modelhub = ModelHub.from_github_repo()
+modelhub.parse_objects()
+
+db_dir = Path(__file__).parent.parent.joinpath("database2309")
+input_dirname = "click_on_the_star_with_a_texture_of_bricks_default"
+
+# Select model
+model_name = "star_with_a_texture_of_bricks_2309_yolov8s-seg.onnx"
+classes = ["star-bricks"]
+
+
+def yolov8_segment(images_dir: Path, output_dir: Path):
+    session = modelhub.match_net(model_name)
+    yoloseg = YOLOv8Seg.from_pluggable_model(session, classes)
+
+    # Initialize progress bar
+    desc_in = f'"{images_dir.parent.name}/{images_dir.name}"'
+    with tqdm(total=len(os.listdir(images_dir)), desc=f"Labeling | {desc_in}") as progress:
+        for image_name in os.listdir(images_dir):
+            image_path = images_dir.joinpath(image_name)
+            if not image_path.is_file():
+                progress.total -= 1
+                continue
+            # Find all the circles in the picture
+            yoloseg(image_path, shape_type="point")
+
+            # Draw a bounding box and mask region for all circles
+            img = cv2.imread(str(image_path))
+            combined_img = yoloseg.draw_masks(img, mask_alpha=0.5)
+            output_path = output_dir.joinpath(image_path.name)
+            cv2.imwrite(str(output_path), combined_img)
+
+            progress.update(1)
+
+    if "win32" in sys.platform and "PYTEST_RUN_CONFIG" not in os.environ:
+        os.startfile(output_dir)
+    print(f">> View at {output_dir}")
+
+
+def demo():
+    images_dir = db_dir.joinpath(input_dirname)
+
+    output_dir = Path(__file__).parent.joinpath(
+        "yolo_mocker", "figs-star-bricks-seg-out", images_dir.name
+    )
+    shutil.rmtree(output_dir, ignore_errors=True)
+    output_dir.mkdir(parents=True, exist_ok=True)
+
+    yolov8_segment(images_dir, output_dir)
+
+
+if __name__ == "__main__":
+    demo()

automation/flow_card.py

@@ -21,4 +21,9 @@
         "negative_labels": ["bicycle", "off-road vehicle"],
         "joined_dirs": ["sedan_car"],
     },
+    {
+        "positive_labels": ["turtle"],
+        "negative_labels": ["horse", "bear", "giraffe", "dolphins"],
+        "joined_dirs": ["please_click_on_the_smallest_animal", "nested_smallest_turtle"],
+    },
 ]

automation/mini_workflow.py

@@ -88,11 +88,12 @@ def upgrade_objects(aid_):
     # fmt:off
     focus_flags = {
         # "<diagnosed_label_name>": "<model_name[flag]>"
-        "sedan_car": "sedan_car2309",
+        # "sedan_car": "sedan_car2309",
+        "nested_smallest_turtle": "nested_smallest_turtle2309",
     }
     # fmt:on
 
     quick_train()
     aid = quick_development()
-    upgrade_objects(aid)
+    # upgrade_objects(aid)
     print(aid)