Yolo-Pest: An Insect Pest Object Detection Algorithm via CAC3 Module

🎉Yolo-Pest finished the game with a leading mAP.5 of 91.9% on 10th Teddy Cup Question A, wining the national first prize.🎉

📖How to use

📚Dataset

Download dataset from the official website of the Teddy Cup (including training image of 28 species of pests and label data).

Download IP102 from IP102 (including more than 75,000 training image of 102 kinds of pests and label data).

After downloading the dataset, the format of the labels needs to be processed into the yolo format.

0 0.96501 0.9  0.0733788 0.1
0 0.35401 0.2  0.0045253 0.5
...

You also need a classes.txt in the same folder to display the label name.

Sesamia inferens
Chilo suppressalis
Rice leaf roller
...

After downloading the dataset, check that the path is correct before running.

The structure of the folders in the entire dataset is as follows:

DATAROOT
└── IP102
    ├── images
    │   ├── trains
    │   │   ├──IP000000002.jpg
    │   │   ├──...
    │   ├── vals
    │   │   ├──IP000000000.jpg
    │   │   ├──...
    └──labels
    │   ├── trains
    │   │   ├──IP000000002.txt
    │   │   ├──classes.txt
    │   │   ├──...
    │   ├── vals
    │   │   ├──IP000000000.txt
    │   │   ├──classes.txt
    │   │   ├──...
   ...

Enter the address and label name of your data set in the insect.yaml which ensure that the dataset is used during the training process.

💾Environment

Our environment: Ubuntu 18.04, CUDA 11.2

Use conda to create the conda environment and activate it:

conda env create --name your_env_name python=3.8
conda activate your_env_name
pip install -r requirements.txt

📈Training

Check the path in train.py, and run it to train:

python3 train.py 

You can change the parameter from line 430 in train.py to get an improved performance.

parser.add_argument('--weights', type=str, help='initial weights path')
parser.add_argument('--cfg', type=str, help='model.yaml path')
parser.add_argument('--data', type=str, help='dataset.yaml path')
parser.add_argument('--hyp', type=str, help='hyperparameters path')
parser.add_argument('--epochs', type=int, help='total training epochs')
parser.add_argument('--batch-size', type=int, help='total batch size for all GPUs, -1 for autobatch')
...

📝Validation and Evaluation

Check the path in val.py.

Then, run it to get the predicting results and IoU in paper separately(It will run after each epoch by default during training ):

python3 val.py

👀Visualization

Use tensorboard to visualization

pip install tensorboard
tensorboard --logdir my_log

👏Acknowledgment

This repo is based on Yolov5, and we are very grateful for their excellent work.