Upload folder using huggingface_hub

README.md

@@ -1,152 +1,97 @@
 ---
 license: other
-license_name: bsl-1.1
-license_link: https://mariadb.com/bsl11/
-library_name: ultralytics
+license_name: business-source-license
+license_link: LICENSE
 tags:
-  - onnx
   - yolo
-  - yolov8
   - object-detection
   - whiteboard
   - diagram
-  - shapes
-pipeline_tag: object-detection
+  - flowchart
+  - onnx
+library_name: onnxruntime
 ---
 
-# Whiteboard Detector
-
-**Detects hand-drawn shapes on whiteboards.**
+# Cyan Sketch - Whiteboard Shape Detector
 
-YOLOv8-nano fine-tuned to recognize 30 diagram shape classes.
+YOLOv8n model for detecting shapes and connectors in whiteboard/flowchart images.
 
-## Quick Stats
+## Model Details
 
-| Spec | Value |
-|------|-------|
-| Architecture | YOLOv8-nano |
-| Format | ONNX |
-| Size | ~12 MB |
-| Input | 640×640 RGB |
-| Classes | 30 |
-| Training | 100 epochs, 211 images |
-| Hardware | M3 Max, 1.4 hours |
-
-## Classes (30)
+- **Architecture**: YOLOv8n (nano)
+- **Format**: ONNX
+- **Input Size**: 640x640
+- **Classes**: 30 shape types
 
-```
-rectangle, rounded_rectangle, oval, circle, diamond, hexagon,
-parallelogram, triangle, star, cloud, cylinder, stick_figure,
-arrow_box, document_shape, database_icon, square, ellipse,
-pentagon, cross, heart, lightning, banner, callout, bracket,
-solid_arrow, dashed_arrow, bidirectional_arrow, dotted_line,
-curved_arrow, curved_line
-```
-
-## Usage
+## Performance
 
-### Python (ultralytics)
+| Metric | Value |
+|--------|-------|
+| mAP50 | 0.592 |
+| mAP50-95 | 0.339 |
 
-```python
-from ultralytics import YOLO
+### Per-Class Performance (Top 10)
 
-model = YOLO("best.onnx")
-results = model("whiteboard.jpg")
+| Class | mAP50 |
+|-------|-------|
+| rounded_rectangle | 0.995 |
+| stick_figure | 0.995 |
+| cloud | 0.980 |
+| rectangle | 0.857 |
+| sticky_note | 0.857 |
+| cylinder | 0.823 |
+| text_label | 0.774 |
+| circle | 0.738 |
+| oval | 0.735 |
+| diamond | 0.713 |
 
-for box in results[0].boxes:
-    cls = int(box.cls[0])
-    conf = float(box.conf[0])
-    x1, y1, x2, y2 = box.xyxy[0].tolist()
-    print(f"{model.names[cls]}: {conf:.2f} at ({x1:.0f}, {y1:.0f})")
+## Classes (30)
+```
+rectangle, rounded_rectangle, oval, circle, diamond, triangle,
+cylinder, cloud, hexagon, parallelogram, sticky_note, stick_figure,
+solid_arrow, dashed_arrow, bidirectional_arrow, line, curved_arrow,
+start_dot, end_dot, text_label, ellipse, square,
+curved_bidirectional_arrow, dashed_line, dotted_line, dotted_arrow,
+solid_circle, double_solid_line, dashed_oval, curved_line
 ```
 
-### Python (onnxruntime)
-
+## Usage
 ```python
 import onnxruntime as ort
+import cv2
 import numpy as np
-from PIL import Image
 
 # Load model
 session = ort.InferenceSession("best.onnx")
 
-# Preprocess
-img = Image.open("whiteboard.jpg").resize((640, 640))
-input_tensor = np.array(img).transpose(2, 0, 1).astype(np.float32) / 255.0
-input_tensor = input_tensor[np.newaxis, ...]
-
-# Inference
-outputs = session.run(None, {"images": input_tensor})
-
-# outputs[0] shape: [1, 34, 8400]
-# 34 = 4 (xywh) + 30 (class scores)
-# 8400 = detection candidates
+# Load classes
+with open("classes.txt") as f:
+    classes = [l.strip() for l in f]
+
+# Preprocess image
+img = cv2.imread("whiteboard.jpg")
+resized = cv2.resize(img, (640, 640))
+blob = cv2.cvtColor(resized, cv2.COLOR_BGR2RGB).astype(np.float32) / 255.0
+blob = np.transpose(blob, (2, 0, 1))[None, ...]
+
+# Run inference
+outputs = session.run(None, {"images": blob})[0]
+
+# Parse detections (conf > 0.3)
+for i in range(outputs.shape[2]):
+    scores = outputs[0, 4:, i]
+    class_id = np.argmax(scores)
+    conf = scores[class_id]
+    if conf > 0.3:
+        print(f"{classes[class_id]}: {conf:.2f}")
 ```
 
-### CLI (ultralytics)
-
-```bash
-yolo predict model=best.onnx source=whiteboard.jpg
-```
-
-## Output Format
-
-YOLO outputs tensor `[1, 34, 8400]`:
-
-```
-For each of 8400 candidates:
-  [0] x_center (0-640)
-  [1] y_center (0-640)
-  [2] width
-  [3] height
-  [4-33] confidence per class (30 classes)
-```
-
-Post-process with confidence threshold (0.25) and NMS (0.45 IoU).
-
-## Training Performance
-
-| Class | mAP50 | Notes |
-|-------|-------|-------|
-| cloud | 0.993 | Excellent |
-| rounded_rectangle | 0.995 | Excellent |
-| stick_figure | 0.895 | Good |
-| oval | 0.849 | Good |
-| rectangle | 0.716 | Good |
-| text_label | 0.664 | Fair |
-| solid_arrow | 0.368 | Needs more data |
-| triangle | 0.316 | Needs more data |
-| cylinder | 0.045 | Needs more data |
-
 ## Files
 
-```
-whiteboard-detector/
-├── best.onnx        # Model (use this)
-├── best.pt          # PyTorch weights
-├── classes.txt      # Class names
-├── README.md        # This file
-└── SKILL.md         # Manifest
-```
-
-## Training Data
-
-- 211 annotated whiteboard images
-- Hand-drawn diagrams, varying styles
-- Augmentation: rotation, blur, noise
-
-## Limitations
-
-- Best with clear contrast (dark ink on white)
-- Small shapes (<20px) may be missed
-- Overlapping shapes can confuse detection
-- Some classes undertrained (cylinder, triangle)
+- `best.onnx` - ONNX model (6MB)
+- `classes.txt` - Class names
+- `ocr_dictionary.json` - Domain terms for OCR correction
 
 ## License
 
-**Business Source License 1.1 (BSL-1.1)**
-
-Copyright (c) 2024 Block Xaero Inc.
-
-- ✅ Free for non-production use
-- ⚠️ Production use requires license
+Business Source License - See LICENSE file