YOLOv8 Object Detection & Image Segmentation Implementation (Easy Steps)
- Object Detection
- Image Segmentation
- Document
- Download the pretrained YOLOv8 weights or you can use your own custom trained weights, and paste it in the main folder.
| Object Detection Models |
|---|
| YOLOv8n | | YOLOv8s | | YOLOv8m | | YOLOv8l | | YOLOv8x |
| Image Segmentation Models |
|---|
| YOLOv8n-seg | | YOLOv8s-seg | | YOLOv8m-seg | | YOLOv8l-seg | | YOLOv8x-seg |
- Create Conda Environment
conda create –n yolov8 python=3.9
conda activate yolov8
- Pip install the ultralytics package including all requirements.txt
pip install ultralytics
- Run yolov8 directly on Command Line Interface (CLI) with commands mentioned below. It has various hyperparameters and configurations.
yolo task=detect mode=predict model=yolov8n.pt source=img.jpg #object detection on image
yolo task=detect mode=predict model=yolov8n.pt source=1.jpg conf=0.5 # Set the confidence level at 0.5
yolo task=detect mode=predict model=yolov8n.pt source=1.jpg conf=0.5 show=true # Show output in real-time
yolo task=detect mode=predict model=yolov8n.pt source=1.jpg conf=0.5 save_txt=true # Save the bounding boxes information
yolo task=detect mode=predict model=yolov8n.pt source=1.jpg conf=0.5 save_crop=true # Save cropped objects
yolo task=detect mode=predict model=yolov8n.pt source=1.jpg conf=0.5 save_crop=true hide_labels=true hide_conf=true #Remove the label and confidence level
yolo task=detect mode=predict model=yolov8s.pt source=0 #Object Detection on webcam
yolo task=detect mode=predict model=yolov8s.pt source=video.mp4 show=true # Object Detection on MP4 Video
yolo task=detect mode=predict model=yolov8s.pt source='C:\Users\zeeshan\Desktop\yolov8' #Object Detection on directory
- Use any model and source just uncomment it and run file on conda environment.
from ultralytics import YOLO
#import cv2
#import time
#import os
# Object Detection Models
#model = YOLO("yolov8x.pt") # Detection ( Extra Large )
#model = YOLO("yolov8l.pt") # Detection ( Large Model )
#model = YOLO("yolov8m.pt") # Detection ( Medium Model)
#model = YOLO("yolov8s.pt") # Detection ( Small Model )
model = YOLO("yolov8n.pt") # Detection ( Nano Model )
#Segmentation Model
#model = YOLO("yolov8m-seg.pt") # Segmentation (Medium Model)
# Predictions for Directory Folder, videos, images, and webcam
model.predict(source="C:/Users/zeeshan/Downloads/yolov8/Data/Images", show=True, save=True) # Images Directory Folder
#model.predict(source="C:/Users/zeeshan/Desktop/yolov8/Data/Videos", show=True, save=True) # Videos Directory Folder
#model.predict(source= '0', show=True, save=True) # Webcam
#model.info(verbose=True) ' # Print model information
#model.export(format="onnx") #export model into ONXX
