Update README.md

README.md

@@ -1,79 +1,67 @@
 
-# Bottle Detection with Ultralytics YOLO
 
-This project provides scripts and instructions to train and run inference on a bottle (or bottle cap) detection dataset using Ultralytics YOLO (v8+).
+# Quick Inference with Ultralytics YOLO
 
-## Environment Setup
-
-1. **Install dependencies** (recommended: use a virtual environment):
-	 ```bash
-	 python3 -m venv .venv
-	 source .venv/bin/activate
-	 pip install ultralytics
-	 ```
-
-2. **Prepare your dataset**
-	 - Organize your dataset as:
-		 ```
-		 dataset/
-			 images/
-				 train/
-				 val/
-			 labels/
-				 train/
-				 val/
-			 data.yaml
-		 ```
-	 - The `data.yaml` should specify absolute paths for `train` and `val` images.
-
-## Training a Model
-
-Edit and run `train.py` to train a model from scratch or fine-tune a pretrained model:
+This guide shows how to load a trained or pretrained YOLO model and run inference, returning the center coordinates of detected objects for class 0 and 1.
 
-```
-from ultralytics import YOLO
-
-model = YOLO('yolov8n.pt')  # or your custom/pretrained .pt file
+## Environment Setup
 
-model.train(
-		data='/absolute/path/to/data.yaml',
-		epochs=50,
-		imgsz=640,
-		batch=16,
-		project='runs/train',
-		name='experiment_name',
-)
+```bash
+python3 -m venv .venv
+source .venv/bin/activate
+pip install ultralytics
 ```
 
-**Tips:**
-- You can adjust `epochs`, `imgsz`, `batch`, and other parameters as needed.
-- For small objects, consider increasing `imgsz` (e.g., 1024).
-- Use a GPU for best performance.
+## Inference Example
 
-## Inference with a Trained Model
-
-Edit and run `infer.py` to run inference on images:
-
-```
+```python
 from ultralytics import YOLO
 
-model = YOLO('/absolute/path/to/weights/best.pt')
+# 1. Load your model
+model = YOLO("/home/vmo/Workspace/bottle_detection/runs/detect/runs/train/yolo11n_bottle/weights/best.pt")
 
+# 2. Run prediction
 results = model.predict(
-		source='/path/to/image/or/folder',
-		save=True,           # Save images with boxes
-		conf=0.15,           # Confidence threshold
-		classes=[0, 1]       # (Optional) restrict to certain classes
+    source="/home/vmo/Workspace/bottle_detection/raw_data_test/ros_4_25_45/images/frame_000104.jpg", 
+    save=False, 
+    conf=0.15, 
+    classes=[0, 1]
 )
-```
 
-Output images with bounding boxes will be saved in a `runs/predict` directory.
-
-## Using Pretrained Models
+# 3. Extract Center Coordinates
+# We use a dictionary to store lists of centers in case there are multiple bottles/caps
+centers = {0: [], 1: []} 
+
+for r in results:
+    # .xywh returns [x_center, y_center, width, height] in pixels
+    # .cls returns the class index (0 or 1)
+    boxes = r.boxes.xywh.cpu().numpy()  
+    clss = r.boxes.cls.cpu().numpy()    
+
+    for i, box in enumerate(boxes):
+        x_c, y_c, w, h = box
+        class_id = int(clss[i])
+        centers[class_id].append((float(x_c), float(y_c)))
+
+# 4. Display or use the coordinates
+bottle_centers = centers[0]
+cap_centers = centers[1]
+
+print(f"Bottle Centers (Class 0): {bottle_centers}")
+print(f"Cap Centers (Class 1): {cap_centers}")
+
+# Example: If you need the first detection for a calculation
+if bottle_centers and cap_centers:
+    b_x, b_y = bottle_centers[0]
+    c_x, c_y = cap_centers[0]
+    print(f"Primary Bottle Center: x={b_x:.2f}, y={b_y:.2f}")
+```
 
-You can use any YOLOv8 pretrained weights (e.g., `yolov8n.pt`, `yolov8s.pt`) as the starting point for training or inference by changing the model path in the scripts.
+## Notes
+- Replace `/absolute/path/to/weights/best.pt` with your trained or pretrained model path.
+- Replace `/path/to/image/or/folder` with your image or folder path.
+- The function `get_centers` returns a dictionary with lists of center coordinates for class 0 and 1.
 
 ## References
-
 - [Ultralytics YOLO Docs](https://docs.ultralytics.com/)
 - [YOLOv8 GitHub](https://github.com/ultralytics/ultralytics)