Hi Tim
Finally a few real world early snaps after getting the “camera” all field-ready, running headless on a battery pack etc. Focussing the camera was an early hurdle as I found it too hard to guarantee focus with old eyes and just a wee 2.8” LCD screen. The depth of field on the 35mm lens (equivalent to ~160mm on a full size camera) I am using is quite narrow. I found using RPi-cam-web-interface (https://github.com/silvanmelchior/RPi_Cam_Web_Interface), which I already had installed, works really well as I can focus the camera using my much bigger iPad screen via the web interface. I can then go inside and do a double-check by downloading and looking at a full (5k) hi-res still image on my Mac, again using RPi-cam-web-interface. If its ok I can then shutdown the RPi-cam-web-interface using ssh and startup my birdcam setup with a shell script that shuttles to and from a customised version of your python object detection code. I have now also sorted out the scp commands to transfer the image files off the rpi to my Mac. Storing the images on the SD card gives a much faster (about 3x faster) raspistill image burst rate than saving to a usb drive so less opportunities lost. You quickly see that bird movements are often very quick, even when not flying, and sometimes their appearances at any one spot are fleeting.
I’m pretty pleased with the results so far. The quality of the HQ camera seems amazing to me although I’m no camera aficionado. I’m now waiting to see if sports mode on the camera improves some of the images. The weather hasn’t been conducive to leaving my RPi camera outdoors the last few days. I don’t think it would appreciate a wash.
