ultralytics 8.0.183 RayTune and yolo checks fixes (#5002)

Co-authored-by: Kapil Raj <103250862+raj-kapil@users.noreply.github.com>
Co-authored-by: Muhammad Rizwan Munawar <62513924+RizwanMunawar@users.noreply.github.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

docs/modes/train.md

@@ -189,7 +189,7 @@ Training settings for YOLO models refer to the various hyperparameters and confi
 | `project`         | `None`   | project name                                                                                   |
 | `name`            | `None`   | experiment name                                                                                |
 | `exist_ok`        | `False`  | whether to overwrite existing experiment                                                       |
-| `pretrained`      | `True`   | (bool | str) whether to use a pretrained model (bool) or a model to load weights from (str)    |
+| `pretrained`      | `True`   | (bool \| str) whether to use a pretrained model (bool) or a model to load weights from (str)   |
 | `optimizer`       | `'auto'` | optimizer to use, choices=[SGD, Adam, Adamax, AdamW, NAdam, RAdam, RMSProp, auto]              |
 | `verbose`         | `False`  | whether to print verbose output                                                                |
 | `seed`            | `0`      | random seed for reproducibility                                                                |
@@ -202,7 +202,7 @@ Training settings for YOLO models refer to the various hyperparameters and confi
 | `amp`             | `True`   | Automatic Mixed Precision (AMP) training, choices=[True, False]                                |
 | `fraction`        | `1.0`    | dataset fraction to train on (default is 1.0, all images in train set)                         |
 | `profile`         | `False`  | profile ONNX and TensorRT speeds during training for loggers                                   |
-| `freeze`          | `None`   | (int or list, optional) freeze first n layers, or freeze list of layer indices during training |
+| `freeze`          | `None`   | (int \| list, optional) freeze first n layers, or freeze list of layer indices during training |
 | `lr0`             | `0.01`   | initial learning rate (i.e. SGD=1E-2, Adam=1E-3)                                               |
 | `lrf`             | `0.01`   | final learning rate (lr0 * lrf)                                                                |
 | `momentum`        | `0.937`  | SGD momentum/Adam beta1                                                                        |

docs/usage/cfg.md

@@ -88,7 +88,7 @@ The training settings for YOLO models encompass various hyperparameters and conf
 | `project`         | `None`   | project name                                                                                   |
 | `name`            | `None`   | experiment name                                                                                |
 | `exist_ok`        | `False`  | whether to overwrite existing experiment                                                       |
-| `pretrained`      | `True`   | (bool | str) whether to use a pretrained model (bool) or a model to load weights from (str)    |
+| `pretrained`      | `True`   | (bool \| str) whether to use a pretrained model (bool) or a model to load weights from (str)   |
 | `optimizer`       | `'auto'` | optimizer to use, choices=[SGD, Adam, Adamax, AdamW, NAdam, RAdam, RMSProp, auto]              |
 | `verbose`         | `False`  | whether to print verbose output                                                                |
 | `seed`            | `0`      | random seed for reproducibility                                                                |
@@ -101,7 +101,7 @@ The training settings for YOLO models encompass various hyperparameters and conf
 | `amp`             | `True`   | Automatic Mixed Precision (AMP) training, choices=[True, False]                                |
 | `fraction`        | `1.0`    | dataset fraction to train on (default is 1.0, all images in train set)                         |
 | `profile`         | `False`  | profile ONNX and TensorRT speeds during training for loggers                                   |
-| `freeze`          | `None`   | (int or list, optional) freeze first n layers, or freeze list of layer indices during training |
+| `freeze`          | `None`   | (int \| list, optional) freeze first n layers, or freeze list of layer indices during training |
 | `lr0`             | `0.01`   | initial learning rate (i.e. SGD=1E-2, Adam=1E-3)                                               |
 | `lrf`             | `0.01`   | final learning rate (lr0 * lrf)                                                                |
 | `momentum`        | `0.937`  | SGD momentum/Adam beta1                                                                        |

examples/YOLOv8-Region-Counter/readme.md

@@ -5,8 +5,8 @@
 
 <div>
   <p align="center">
-  <img src="https://github.com/RizwanMunawar/ultralytics/assets/62513924/978c8dd4-936d-468e-b41e-1046741ec323" width="45%"/>
+  <img src="https://github.com/RizwanMunawar/ultralytics/assets/62513924/5ab3bbd7-fd12-4849-928e-5f294d6c3fcf" width="45%"/>
-    <img src="https://github.com/RizwanMunawar/ultralytics/assets/62513924/069fd81b-8451-40f3-9f14-709a7ac097ca" width="45%"/>
+    <img src="https://github.com/RizwanMunawar/ultralytics/assets/62513924/e7c1aea7-474d-4d78-8d48-b50854ffe1ca" width="45%"/>
 &nbsp; &nbsp; &nbsp; &nbsp;
 </p>
 </div>
@@ -42,6 +42,9 @@ After the video begins playing, you can freely move the region anywhere within t
 # If you want to save results
 python yolov8_region_counter.py --source "path/to/video.mp4" --save-img --view-img
 
+# If you want to run model on CPU
+python yolov8_region_counter.py --source "path/to/video.mp4" --save-img --view-img --device cpu
+
 # If you want to change model file
 python yolov8_region_counter.py --source "path/to/video.mp4" --save-img --weights "path/to/model.pt"
 
@@ -52,10 +55,12 @@ python yolov8_region_counter.py --source "path/to/video.mp4" --view-img
 ## Usage Options
 
 - `--source`: Specifies the path to the video file you want to run inference on.
+- `--device`: Specifies the device `cpu` or `0`
 - `--save-img`: Flag to save the detection results as images.
 - `--weights`: Specifies a different YOLOv8 model file (e.g., `yolov8n.pt`, `yolov8s.pt`, `yolov8m.pt`, `yolov8l.pt`, `yolov8x.pt`).
 - `--line-thickness`: Specifies the bounding box thickness
-- `--region-thickness`: Specific the region boxes thickness
+- `--region-thickness`: Specifies the region boxes thickness
+- `--track-thickness`: Specifies the track line thickness
 
 ## FAQ
 
@@ -63,11 +68,40 @@ python yolov8_region_counter.py --source "path/to/video.mp4" --view-img
 
 Region counting is a computational method utilized to ascertain the quantity of objects within a specific area in recorded video or real-time streams. This technique finds frequent application in image processing, computer vision, and pattern recognition, facilitating the analysis and segmentation of objects or features based on their spatial relationships.
 
-**2. Why Combine Region Counting with YOLOv8?**
+**2. Is Friendly Region Plotting Supported by the Region Counter?**
+
+The Region Counter offers the capability to create regions in various formats, such as polygons and rectangles. You have the flexibility to modify region attributes, including coordinates, colors, and other details, as demonstrated in the following code:
+
+```python
+counting_regions = [
+    {
+        "name": "YOLOv8 Polygon Region",
+        "polygon": Polygon(
+            [(50, 80), (250, 20), (450, 80), (400, 350), (100, 350)]
+        ),  # Polygon with five points (Pentagon)
+        "counts": 0,
+        "dragging": False,
+        "region_color": (255, 42, 4),  # BGR Value
+        "text_color": (255, 255, 255),  # Region Text Color
+    },
+    {
+        "name": "YOLOv8 Rectangle Region",
+        "polygon": Polygon(
+            [(200, 250), (440, 250), (440, 550), (200, 550)]
+        ),  # Rectangle with four points
+        "counts": 0,
+        "dragging": False,
+        "region_color": (37, 255, 225),  # BGR Value
+        "text_color": (0, 0, 0),  # Region Text Color
+    },
+]
+```
+
+**3. Why Combine Region Counting with YOLOv8?**
 
 YOLOv8 specializes in the detection and tracking of objects in video streams. Region counting complements this by enabling object counting within designated areas, making it a valuable application of YOLOv8.
 
-**3. How Can I Troubleshoot Issues?**
+**4. How Can I Troubleshoot Issues?**
 
 To gain more insights during inference, you can include the `--debug` flag in your command:
 
@@ -75,10 +109,10 @@ To gain more insights during inference, you can include the `--debug` flag in yo
 python yolov8_region_counter.py --source "path to video file" --debug
 ```
 
-**4. Can I Employ Other YOLO Versions?**
+**5. Can I Employ Other YOLO Versions?**
 
 Certainly, you have the flexibility to specify different YOLO model weights using the `--weights` option.
 
-**5. Where Can I Access Additional Information?**
+**6. Where Can I Access Additional Information?**
 
 For a comprehensive guide on using YOLOv8 with Object Tracking, please refer to [Multi-Object Tracking with Ultralytics YOLO](https://docs.ultralytics.com/modes/track/).

examples/YOLOv8-Region-Counter/yolov8_region_counter.py

@@ -4,34 +4,37 @@ from pathlib import Path
 
 import cv2
 import numpy as np
+from shapely.geometry import Polygon
+from shapely.geometry.point import Point
 
 from ultralytics import YOLO
-
-track_history = defaultdict(lambda: [])
-
 from ultralytics.utils.files import increment_path
 from ultralytics.utils.plotting import Annotator, colors
 
-# Region utils
+track_history = defaultdict(lambda: [])
+
 current_region = None
-counting_regions = [{
+counting_regions = [
-    'name': 'YOLOv8 Region A',
+    {
-    'roi': (50, 100, 240, 300),
+        'name': 'YOLOv8 Polygon Region',
-    'counts': 0,
+        'polygon': Polygon([(50, 80), (250, 20), (450, 80), (400, 350), (100, 350)]),  # Polygon points
-    'dragging': False,
-    'region_color': (0, 255, 0)}, {
-        'name': 'YOLOv8 Region B',
-        'roi': (200, 250, 240, 300),
         'counts': 0,
         'dragging': False,
-        'region_color': (255, 144, 31)}]
+        'region_color': (255, 42, 4),  # BGR Value
+        'text_color': (255, 255, 255)  # Region Text Color
+    },
+    {
+        'name': 'YOLOv8 Rectangle Region',
+        'polygon': Polygon([(200, 250), (440, 250), (440, 550), (200, 550)]),  # Polygon points
+        'counts': 0,
+        'dragging': False,
+        'region_color': (37, 255, 225),  # BGR Value
+        'text_color': (0, 0, 0),  # Region Text Color
+    }, ]
 
 
-def is_inside_roi(box, roi):
+def is_inside_polygon(point, polygon):
-    """Compare bbox with region box."""
+    return polygon.contains(Point(point))
-    x, y, _, _ = box
-    roi_x, roi_y, roi_w, roi_h = roi
-    return roi_x < x < roi_x + roi_w and roi_y < y < roi_y + roi_h
 
 
 def mouse_callback(event, x, y, flags, param):
@@ -41,18 +44,21 @@ def mouse_callback(event, x, y, flags, param):
     # Mouse left button down event
     if event == cv2.EVENT_LBUTTONDOWN:
         for region in counting_regions:
-            roi_x, roi_y, roi_w, roi_h = region['roi']
+            if is_inside_polygon((x, y), region['polygon']):
-            if roi_x < x < roi_x + roi_w and roi_y < y < roi_y + roi_h:
                 current_region = region
                 current_region['dragging'] = True
-                current_region['offset_x'] = x - roi_x
+                current_region['offset_x'] = x
-                current_region['offset_y'] = y - roi_y
+                current_region['offset_y'] = y
 
     # Mouse move event
     elif event == cv2.EVENT_MOUSEMOVE:
         if current_region is not None and current_region['dragging']:
-            current_region['roi'] = (x - current_region['offset_x'], y - current_region['offset_y'],
+            dx = x - current_region['offset_x']
-                                     current_region['roi'][2], current_region['roi'][3])
+            dy = y - current_region['offset_y']
+            current_region['polygon'] = Polygon([
+                (p[0] + dx, p[1] + dy) for p in current_region['polygon'].exterior.coords])
+            current_region['offset_x'] = x
+            current_region['offset_y'] = y
 
     # Mouse left button up event
     elif event == cv2.EVENT_LBUTTONUP:
@@ -60,26 +66,33 @@ def mouse_callback(event, x, y, flags, param):
             current_region['dragging'] = False
 
 
-def run(weights='yolov8n.pt',
+def run(
-        source='test.mp4',
+    weights='yolov8n.pt',
-        view_img=False,
+    source=None,
-        save_img=False,
+    device='cpu',
-        exist_ok=False,
+    view_img=False,
-        line_thickness=2,
+    save_img=False,
-        region_thickness=2):
+    exist_ok=False,
+    line_thickness=2,
+    track_thickness=2,
+    region_thickness=2,
+):
     """
     Run Region counting on a video using YOLOv8 and ByteTrack.
 
     Supports movable region for real time counting inside specific area.
     Supports multiple regions counting.
+    Regions can be Polygons or rectangle in shape
 
     Args:
         weights (str): Model weights path.
         source (str): Video file path.
+        device (str): processing device cpu, 0, 1
         view_img (bool): Show results.
         save_img (bool): Save results.
         exist_ok (bool): Overwrite existing files.
         line_thickness (int): Bounding box thickness.
+        track_thickness (int): Tracking line thickness
         region_thickness (int): Region thickness.
     """
     vid_frame_count = 0
@@ -90,6 +103,7 @@ def run(weights='yolov8n.pt',
 
     # Setup Model
     model = YOLO(f'{weights}')
+    model.to('cuda') if device == '0' else model.to('cpu')
 
     # Video setup
     videocapture = cv2.VideoCapture(source)
@@ -122,40 +136,43 @@ def run(weights='yolov8n.pt',
             label = str(names[cls])
             xyxy = (x - w / 2), (y - h / 2), (x + w / 2), (y + h / 2)
 
-            # Bounding box
+            # Bounding box plot
             bbox_color = colors(cls, True)
             annotator.box_label(xyxy, label, color=bbox_color)
 
-            # Tracking Lines
+            # Tracking Lines plot
             track = track_history[track_id]
             track.append((float(x), float(y)))
             if len(track) > 30:
                 track.pop(0)
             points = np.hstack(track).astype(np.int32).reshape((-1, 1, 2))
-            cv2.polylines(frame, [points], isClosed=False, color=bbox_color, thickness=line_thickness)
+            cv2.polylines(frame, [points], isClosed=False, color=bbox_color, thickness=track_thickness)
 
-            # Check If detection inside region
+            # Check if detection inside region
             for region in counting_regions:
-                if is_inside_roi(box, region['roi']):
+                if is_inside_polygon((x, y), region['polygon']):
                     region['counts'] += 1
 
-        # Draw region boxes
+        # Draw regions (Polygons/Rectangles)
         for region in counting_regions:
             region_label = str(region['counts'])
-            roi_x, roi_y, roi_w, roi_h = region['roi']
             region_color = region['region_color']
-            center_x = roi_x + roi_w // 2
+            region_text_color = region['text_color']
-            center_y = roi_y + roi_h // 2
-            text_margin = 15
 
-            # Region plotting
+            polygon_coords = np.array(region['polygon'].exterior.coords, dtype=np.int32)
-            cv2.rectangle(frame, (roi_x, roi_y), (roi_x + roi_w, roi_y + roi_h), region_color, region_thickness)
+            centroid_x, centroid_y = int(region['polygon'].centroid.x), int(region['polygon'].centroid.y)
-            t_size, _ = cv2.getTextSize(region_label, cv2.FONT_HERSHEY_SIMPLEX, fontScale=1.0, thickness=line_thickness)
+
-            text_x = center_x - t_size[0] // 2 - text_margin
+            text_size, _ = cv2.getTextSize(region_label,
-            text_y = center_y + t_size[1] // 2 + text_margin
+                                           cv2.FONT_HERSHEY_SIMPLEX,
-            cv2.rectangle(frame, (text_x - text_margin, text_y - t_size[1] - text_margin),
+                                           fontScale=0.7,
-                          (text_x + t_size[0] + text_margin, text_y + text_margin), region_color, -1)
+                                           thickness=line_thickness)
-            cv2.putText(frame, region_label, (text_x, text_y), cv2.FONT_HERSHEY_SIMPLEX, 1.0, (0, 0, 0), line_thickness)
+            text_x = centroid_x - text_size[0] // 2
+            text_y = centroid_y + text_size[1] // 2
+            cv2.rectangle(frame, (text_x - 5, text_y - text_size[1] - 5), (text_x + text_size[0] + 5, text_y + 5),
+                          region_color, -1)
+            cv2.putText(frame, region_label, (text_x, text_y), cv2.FONT_HERSHEY_SIMPLEX, 0.7, region_text_color,
+                        line_thickness)
+            cv2.polylines(frame, [polygon_coords], isClosed=True, color=region_color, thickness=region_thickness)
 
         if view_img:
             if vid_frame_count == 1:
@@ -182,12 +199,15 @@ def parse_opt():
     """Parse command line arguments."""
     parser = argparse.ArgumentParser()
     parser.add_argument('--weights', type=str, default='yolov8n.pt', help='initial weights path')
+    parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
     parser.add_argument('--source', type=str, required=True, help='video file path')
     parser.add_argument('--view-img', action='store_true', help='show results')
     parser.add_argument('--save-img', action='store_true', help='save results')
     parser.add_argument('--exist-ok', action='store_true', help='existing project/name ok, do not increment')
     parser.add_argument('--line-thickness', type=int, default=2, help='bounding box thickness')
+    parser.add_argument('--track-thickness', type=int, default=2, help='Tracking line thickness')
     parser.add_argument('--region-thickness', type=int, default=4, help='Region thickness')
+
     return parser.parse_args()
 
 

ultralytics/__init__.py

@@ -1,6 +1,6 @@
 # Ultralytics YOLO 🚀, AGPL-3.0 license
 
-__version__ = '8.0.182'
+__version__ = '8.0.183'
 
 from ultralytics.models import RTDETR, SAM, YOLO
 from ultralytics.models.fastsam import FastSAM

ultralytics/utils/checks.py

@@ -493,9 +493,15 @@ def collect_system_info():
                 f"{'CPU':<20}{get_cpu_info()}\n"
                 f"{'CUDA':<20}{torch.version.cuda if torch and torch.cuda.is_available() else None}\n")
 
-    for r in parse_requirements():
+    if (ROOT.parent / 'requirements.txt').exists():  # pip install
+        requirements = parse_requirements()
+    else:  # git install
+        from pkg_resources import get_distribution
+        requirements = get_distribution('ultralytics').requires()
+
+    for r in requirements:
         current = version(r.name)
-        is_met = '✅ ' if check_version(current, r.specifier) else '❌ '
+        is_met = '✅ ' if check_version(current, str(r.specifier)) else '❌ '
         LOGGER.info(f'{r.name:<20}{is_met}{current}{r.specifier}')
 
 

ultralytics/utils/tuner.py

@@ -124,7 +124,7 @@ def run_ray_tune(model,
     tuner_callbacks = [WandbLoggerCallback(project='YOLOv8-tune')] if wandb else []
 
     # Create the Ray Tune hyperparameter search tuner
-    tune_dir = get_save_dir(DEFAULT_CFG, name='tune')
+    tune_dir = get_save_dir(DEFAULT_CFG, name='tune').resolve()  # must be absolute dir
     tune_dir.mkdir(parents=True, exist_ok=True)
     tuner = tune.Tuner(trainable_with_resources,
                        param_space=space,