SMOKE_CLASS_DETECTION

Table of content

• Install LIBRAIRIES
• Load Model Yolov5
• Dataset
• Training
• Performances
• Results

Install LIBRAIRIES

Clone repo and install requirements.txt in a Python>=3.7.0 environment, including PyTorch>=1.7

git clone https://github.com/ultralytics/yolov5  # clone
cd yolov5
pip install -r requirements.txt  # install

Load Model Yolov5

Models download automatically from the latest YOLOv5 release

import torch

# Model
model = torch.hub.load('ultralytics/yolov5', 'yolov5s')  # or yolov5n - yolov5x6, custom

# Images
img = 'https://img.freepik.com/photos-gratuite/fumee-industrielle-dans-atmosphere_33799-3042.jpg?w=1380'  # or file, Path, PIL, OpenCV, numpy, list

# Inference
results = model(img)

# Results
results.print()  # or .show(), .save(), .crop(), .pandas(), etc.

Dataset

we imported 12336 unlabeled images from kaggle, we labeled 100 images using two classes: 'smoke low density' and 'smoke high density' The same database we found already labeled, with a single class 'smoke'.

Training

We installed wandb to visualize the results directly on the website by creating an account

%pip install -q wandb
import wandb
wandb.login()

The commands below reproduce YOLOv5 data results. Models download automatically from the latest YOLOv5 release

python train.py --data data.yaml --cfg yolov5s.yaml --weights '' --batch-size 64

Performances

Results

The model manages to detect both classes of smoke in this video