Weather Classifier

Image Weather Classification module from DILAM paper & DILAM Github.

This machine learning approach uses the first two layers of YOLOv3 (commit d353371) object detection model, pre-trained on a clear weather dataset, for further image classication of adverse weather conditions.

Installation on Ubuntu 20.04

• Clone the repository recursively:

git clone --recurse-submodules https://github.com/nadezola/weather_classifier.git

• If you already cloned and forgot to use --recurse-submodules you can run:

git submodule update --init

• We recommend to use Python3.8 virtual environment with requirements.txt:

# apt install python3.8-venv
python3.8 -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install -r requirements.txt

Dataset preparation

Data structure:

data_root
|
├── images
|      ├──── test
|      |      ├─── *.jpg
|      |
|      ├──── train
|      |      ├─── *.jpg
|      |
|      └──── val
|             ├─── *.jpg
|
├── splits
|      ├──── clear
|      |      ├─── train.txt
|      |      └─── val.txt
|      |
|      ├──── fog
|      |      ├─── train.txt
|      |      └─── val.txt
|      |
|      ├──── rain
|      |      ├─── train.txt
|      |      └─── val.txt
|      |
|      └─── snow
|            ├─── train.txt
|            └─── val.txt
|
├── test.txt
├── train.txt
├── val.txt
|
└── weather_labels.csv 

• test.txt, train.txt, val.txt contain a list of images defining the train/val/test splits.

Splitting by Weather Condition:

• To train the Weather Classifier for 4 classes: clear, fog, rain, snow, you need to split the entire train.txt and val.txt by weather conditions and fill up the folder splits
• We provide example code to split the entire train.txt and val.txt by weather conditions, using weather_labels.csv annotation file
• Example of weather_labels.csv:

Hash	Weather
vienna20181007_f0	fog
vienna20181007_f1	fog
...
vienna20181015_f0	rain

• Usage:

python data_splitting.py --data_root <path/to/data_root> 
                         --weather_lbls <path/to/weather_labels>
                         --res_dir <path/to/folder/where/to/save/results>

Train

Train the Weather Classification Head

• Configure opt.py file
• Pre-train the YOLOv3 model commit d353371 on clear weather dataset for your object detection task and put the pre-trained model in checkpoints folder
• Run:

python main.py --data_root <path/to/data/root/> 
               --mode 'train'
               --res_dir <path/to/folder/where/to/save/results>

Evaluate

Evaluate the Weather Classification performance on validation dataset.

• Configure opt.py file
• Put the weights of Weather Classification Head in checkpoints folder
• Run:

python main.py --data_root <path/to/data/root/> 
               --mode 'eval'
               --weights <path/to/weather/classification/weights>
               --res_dir <path/to/folder/where/to/save/results>
               # --novis option to DO NOT visualize weather classification results

Predict

Play the Weather Classifier on test dataset. No labels needed.

• Configure opt.py file
• Put the weights of Weather Classification Head in checkpoints folder
• Run:

python main.py --data_root <path/to/data/root/> 
               --mode 'predict'
               --weights <path/to/weather/classification/weights>
               --res_dir <path/to/folder/where/to/save/results>
               # --novis option to DO NOT visualize weather classification results