Create README.md

README.md

@@ -0,0 +1,189 @@
+Model Description
+
+This model detects bike lane infrastructure and related objects in street images using object detection.
+
+I used a YOLOv11 model, which predicts bounding boxes and class labels for objects such as bike lanes, lane markings, cyclists, and vehicles. The model was fine-tuned from a pre-trained version using transfer learning.
+
+The goal of this project was to understand how well object detection performs in this setting and to evaluate its limitations, rather than just achieving high performance.
+
+Intended Use Cases:
+
+Transportation research
+
+Bike lane detection from street images
+
+Infrastructure analysis
+
+Training Data
+
+Dataset Source:
+Roboflow Universe – Bike Lane Computer Vision Dataset
+
+Dataset Size:
+147 images, 7 classes
+
+Class Distribution:
+
+Class	Count
+Vehicle	253
+Bicycle Lane	129
+Shared Dotted Lane	124
+Solid Lane	59
+Cyclist	13
+Bicycle	2
+Car	2
+
+Data Collection Methodology:
+The dataset consists of urban street images containing bike lanes, vehicles, and cyclists under various lighting and road conditions.
+
+Annotation Process:
+The dataset included pre-existing YOLO bounding box annotations. These annotations label objects using rectangular bounding boxes and class labels.
+
+I reviewed a subset of images to verify:
+
+bounding box alignment
+
+label consistency
+
+No major modifications were made to the annotations. This means that while the dataset was usable, the project relied on existing annotations rather than adding new ones.
+
+Dataset Split:
+
+Train: 102 images (69%)
+
+Validation: 20 images (14%)
+
+Test: 16 images (11%)
+
+Data Augmentation:
+Default YOLO augmentation was used, including flipping and color variation.
+
+Dataset Limitations / Biases:
+
+strong class imbalance
+
+very limited examples for some classes
+
+mostly urban, daytime conditions
+
+Training Procedure
+
+Framework:
+Ultralytics YOLOv11
+
+Training Approach:
+Fine-tuning a pre-trained model
+
+Hardware:
+Google Colab (CPU or GPU — update if needed)
+
+Training Time:
+Approximately ~1 hour
+
+Hyperparameters:
+
+Epochs: 50
+
+Image size: 640
+
+Batch size: 16 (default)
+
+Learning rate: default YOLO setting
+
+Preprocessing:
+
+images resized to 640×640
+
+normalization handled automatically
+
+Evaluation Results
+
+Key Metrics:
+
+Precision: ~0.88
+
+Recall: ~0.38
+
+mAP50: ~0.48
+
+Rather than focusing only on the numbers, these metrics help explain how the model behaves.
+
+The model has relatively high precision, meaning most detections are correct, but lower recall, meaning it misses some objects.
+
+Per-Class Performance
+
+Strong performance on common classes such as vehicles and lane markings
+
+Weak performance on rare classes such as bicycle and car
+
+This is largely due to the imbalance in the dataset.
+
+Visual Examples of Classes
+
+(Upload images showing each class)
+
+Key Visualizations
+
+Include:
+
+confusion matrix
+
+training loss curves
+
+prediction examples
+
+Performance Analysis
+
+The model performs well when:
+
+lane markings are clear
+
+lighting is consistent
+
+The model struggles when:
+
+lane markings are faded or unclear
+
+objects are partially occluded
+
+classes have very few examples
+
+This demonstrates that model performance is strongly influenced by dataset quality and class balance.
+
+Limitations and Biases
+
+Failure Cases:
+
+missed detections of bicycles and cars
+
+errors when lane markings are unclear
+
+confusion between similar lane types
+
+Data Biases:
+
+overrepresentation of vehicles
+
+underrepresentation of rare classes
+
+limited environmental diversity
+
+Environmental Limitations:
+
+poor lighting
+
+occlusion
+
+worn lane markings
+
+Inappropriate Use Cases:
+This model should not be used for:
+
+real-time safety decisions
+
+autonomous driving
+
+high-stakes applications
+
+Sample Size Limitations:
+Classes like bicycle and car have too few examples to be reliably detected.