This is a sample script for determining the image quality sent by DOST-ASTI's UrbanFlood Sensor Networks. DOST-ASTI's project on urban flood monitoring involves a tandem of water-level sensor and serial camera to monitor street level flooding. Data from these sensors are sent via GPRS every 5 minutes.
The challenge in archiving the camera images is that sometimes the images are corrupted upon receiving. To segregate between the good and bad images, a simple machine learning model is made to clean the data set worth of 30000 images.