Object Detection with YOLOv11

Real-time object detection using a webcam, combining a general-purpose YOLO model with a custom-trained adapter/device detector.

What it does

• Live webcam feed — captures frames from your webcam in real time
• General object detection — uses yolo11n.pt to detect everyday objects (people, cups, chairs, etc.)
• Custom device detection — uses a fine-tuned adapter_detector.pt model to detect a specific device/adapter class on top of the general detections
• Both sets of detections are overlaid on the same frame and displayed in a window

Files

File	Description
object_detector.py	Run real-time detection using both models
train.py	Fine-tune yolo11n-seg.pt on the custom adapter dataset
detecting_adapter/data.yaml	Dataset config for the custom training

Setup

python -m venv .venv
source .venv/bin/activate
pip install ultralytics opencv-python

You will also need the model weights (not included in the repo):

• yolo11n.pt — download from Ultralytics
• yolo11n-seg.pt — same source, segmentation variant
• adapter_detector.pt — produced by running train.py

Training the custom model

python train.py

Trains for 100 epochs on the adapter dataset and saves the best weights as adapter_detector.pt.

Running detection

python object_detector.py

Press q to quit.