day_night_classification

Classify your pictures by classes day and night with a CNN implemented with keras and tensorflow as backend!

Predict your images:

python prediction.py --path path/to/your/image/directory

Running the evaluate script results in a .csv-file composed of two columns (filename, classlabel). The pretrained model is used for this step.

Copy them into seperate directories:

python partition.py --path path/to/your/image/directory

Running the partition script results in two directiores in the results dir containing the final partition of the images.

Train the model w/ your own day/night images:

Your train data has to be divided into two directories [../data/day, ../data/night]

python train.py --path path/to/your/train/data

Running the train script results in weights.h5 file and overwrites the previous one.

Authors

• Şiyar Yıkmış

Acknowledgments

• CrowdHuman: A Benchmark for Detecting Human in a Crowd
• A. Pronobis, B. Caputo, P. Jensfelt, and H. I. Christensen. A discriminative approach to robust visual place recognition. In Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS06), Beijing, China, October 2006.