fix some hf util bugs (#33)

* fix some hf util bugs

* update version

* update test

* fix env variable

* fix workflows

* auto download font when not present

* fix font selection

* improve tests

* update tests

* updates

* updates

* updates

* updates

* update

.github/workflows/ci.yml

@@ -63,5 +63,7 @@ jobs:
           pip list
 
       - name: Unittest
+        env:
+          HF_TOKEN: ${{ secrets.HF_TOKEN }}
         run: |
-          pytest
+          pytest -s

.github/workflows/package_ci.yml

@@ -60,5 +60,7 @@ jobs:
           pip list
 
       - name: Unittest
+        env:
+          HF_TOKEN: ${{ secrets.HF_TOKEN }}
         run: |
-          pytest
+          pytest -s

tests/test_cli.py

@@ -3,7 +3,7 @@
 import subprocess
 from pathlib import Path
 
-from ultralytics.yolo.utils import ROOT, SETTINGS
+from ultralytics.yolo.utils import SETTINGS
 
 MODEL = Path(SETTINGS['weights_dir']) / 'yolov8n'
 CFG = 'yolov8n'

tests/test_hf.py

@@ -3,7 +3,10 @@
     download_from_hub,
     postprocess_classify_output,
     render_result,
+    push_to_hfhub,
 )
+import subprocess
+import os
 
 hub_id = "ultralyticsplus/yolov8s"
 
@@ -56,3 +59,36 @@ def test_classification_inference():
     results = model(image, imgsz=640)
     name_to_probs = postprocess_classify_output(model=model, result=results[0])
     name_to_probs
+
+
+def run(cmd):
+    # Run a subprocess command with check=True
+    subprocess.run(cmd.split(), check=True)
+
+
+def test_detection_upload():
+    import platform
+    from packaging.version import Version
+
+    # run following lines if linux and python major == 3 and python minor == 10 (python micor can be anything)
+    print(f'platform.system(): {platform.system()}')
+    print(f'platform.python_version(): {platform.python_version()}')
+    print(f'os.getenv("RUNNER_OS"): {os.getenv("RUNNER_OS")}')
+    print(f'Version(platform.python_version()) >= Version("3.10"): {Version(platform.python_version()) >= Version("3.10")}')
+    if platform.system() == 'Linux' and Version(platform.python_version()) >= Version("3.10"):
+        print('training started')
+        run('yolo train detect model=yolov8n.pt data=coco8.yaml imgsz=32 epochs=1')
+        print('training ended')
+        hf_token = os.getenv('HF_TOKEN')
+        if hf_token is None:
+            raise ValueError('Please set HF_TOKEN environment variable to your HuggingFace token.')
+        print('push to hub started')
+        push_to_hfhub(
+            hf_model_id="fcakyon/yolov8n-test",
+            exp_dir='runs/detect/train',
+            hf_token=os.getenv('HF_TOKEN'),
+            hf_private=True,
+            hf_dataset_id="fcakyon/football-detection",
+            thumbnail_text='YOLOv8s Football Detection'
+        )
+        print('push to hub ended')

ultralyticsplus/__init__.py

@@ -1,4 +1,4 @@
 from .hf_utils import download_from_hub, push_to_hfhub
 from .ultralytics_utils import YOLO, postprocess_classify_output, render_result
 
-__version__ = "0.0.16"
+__version__ = "0.0.17"

ultralyticsplus/hf_utils.py

@@ -204,7 +204,7 @@ def generate_thumbnail(
         text=thumbnail_text,
         pil_image=read_image_as_pil(image_path_or_url),
         brightness=0.60,
-        text_font=65,
+        text_font_size=65,
         crop_margin=None,
     )
 
@@ -508,14 +508,14 @@ def push_to_hfhub(
     score_top5_acc = None
     if task in ["object-detection", "instance-segmentation"]:
         # read the largest value in metrics/mAP50(B) column from csv file named results.csv
-        score_map50 = results_df["metrics/mAP50(B)"].max()
+        score_map50 = results_df["metrics/mAP50(B)"].max().item()
     if task == "instance-segmentation":
         # read the largest value in metrics/mAP50(B) metrics/mAP50(M) columns from csv file named results.csv
-        score_map50_mask = results_df["metrics/mAP50(M)"].max()
+        score_map50_mask = results_df["metrics/mAP50(M)"].max().item()
     if task == "image-classification":
         # read the largest value in metrics/accuracy_top1 metrics/accuracy_top5 columns from csv file named results.csv
-        score_top1_acc = results_df["metrics/accuracy_top1"].max()
-        score_top5_acc = results_df["metrics/accuracy_top5"].max()
+        score_top1_acc = results_df["metrics/accuracy_top1"].max().item()
+        score_top5_acc = results_df["metrics/accuracy_top5"].max().item()
 
     _push_to_hfhub(
         hf_model_id=hf_model_id,

ultralyticsplus/other_utils.py

@@ -1,11 +1,32 @@
+import os
+from PIL.ImageFont import FreeTypeFont
+
+FONT_URL = "https://github.com/fcakyon/ultralyticsplus/releases/download/0.0.16/OpenSans-Bold.ttf"
+
+
+def get_font(size: int) -> FreeTypeFont:
+    from PIL import ImageFont
+
+    try:
+        font = ImageFont.truetype("OpenSans-Bold.ttf", size)
+    except OSError:
+        from sahi.utils.file import download_from_url
+
+        package_dir = os.path.dirname(os.path.abspath(__file__))
+        font_path = os.path.join(package_dir, "OpenSans-Bold.ttf")
+        download_from_url(from_url=FONT_URL, to_path=font_path)
+        font = ImageFont.truetype(font_path, size)
+    return font
+
+
 def add_text_to_image(
     text,
     pil_image,
     brightness: float = 0.75,
-    text_font: int = 50,
+    text_font_size: int = 50,
     crop_margin: int = None,
 ):
-    from PIL import ImageDraw, ImageFont, ImageEnhance
+    from PIL import ImageDraw, ImageEnhance
 
     RESIZE_WIDTH_TO = 1280
     MAX_TEXT_WIDTH = 860
@@ -31,7 +52,7 @@ def add_text_to_image(
     draw = ImageDraw.Draw(pil_image)
 
     # Define the font and text to be written
-    font = ImageFont.truetype("OpenSans-Bold.ttf", text_font)
+    font = get_font(size=text_font_size)
 
     # Get the bounding box of the text
     bbox = draw.textbbox((0, 0), text, font=font)
@@ -42,8 +63,8 @@ def add_text_to_image(
 
     # update font size
     if text_width > MAX_TEXT_WIDTH:
-        text_font = int(MAX_TEXT_WIDTH / text_width * text_font)
-        font = ImageFont.truetype("OpenSans-Bold.ttf", text_font)
+        text_font_size = int(MAX_TEXT_WIDTH / text_width * text_font_size)
+        font = get_font(size=text_font_size)
         bbox = draw.textbbox((0, 0), text, font=font)
         text_width = bbox[2] - bbox[0]
         text_height = bbox[3] - bbox[1]