Although not earth-shattering, it would be interesting to know if common backyard bird feeding habits depend on environmental factors. Knowing this may enable bird-friendly humans the opportunity to provide maximum uninterrupted feeding times within their own backyards.
I propose streaming five data points to Amazon Web Services (AWS) for analysis. I propose using the Intel Galileo to monitor four environmental sensors. (See Sensors section below.) The Galileo provides a means to read sensors and also provides easy access to the AWS.
• Barometric Pressure in millibars, not adjusted for sea level. • Temperature in Centigrade. • Relative Humidity in percentage. • Ambient Light Intensity.
I propose gathering the fifth data point by detecting birds from a photo. A Raspberry PI 3 will control a Nikon D60 digital camera using gphoto2 libraries. Like the Galileo, the Raspberry PI provides easy access to AWS. The interval of the photo capture will be similar to that of the sensor gathering interval. The camera will pointed at my backyard feeder.
I intend to use a third-party object detector presented by PyImageSearch. PyImageSearch uses yet another third-party deep learning algorithm that processes an image against an existing model of numerous objects. The output of this process indicates the existence of individual objects, such as birds.
The number of birds in a given time frame shall be analyzed against one or more of the other four inputs to determine any correlation between the entities.