georoen / MA_Hardware

Using Computer Vision to Count Pedestrians. Hardware Documentation

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Logo

This git is part of the master degree project by Jeroen Staab. As long as the thesis is evaluated, you can find the initial proposal here.

Hardware

Here we use two stardot webcameras. They are boradly used in eviromental outdoor sciences [@ Richardson, Harvard]. Temperature ranges from -40C° to +48C° are no problem [@ Stardot Handbook].

There are two ways to configure the camera. The first is also the recomended one - use a browser to access the web-backend via http. Alternativly you could also access the camera via telnet, SSH is not supported.

Works: FTP upload images to upload@raspberrypi.local via wlan0 from laptop.
Works: FTP upload images to upload@10.42.0.10 via eth0 from stardot.
Works: Use Raspi as Timeserver.
Optional: RGB+IR configuration. See EURAC archiv.zip

TODO: Find best camera position

Works: Static IP
Works: FTP Server

Setup 0 (DEPRECATED)

Stream images from public available webcam. Processing only.

TODO: App script to GIT.
TODO: Strucutre data into new folder per month.

Camera with Internet in field and sending images to RaspberryPi processing images remotely.

Works without Internet. Camera and raspberry placed in weatherproof enclosure, Needs power-supply only. Collect data periodically (SD-Card / FTP in wlan0).

Works: Static IP form stardot + raspi (via eth0, dhcp raspberrypi.local via wlan0).
Works: Configure cameras by port forwaring to localhost:8080 when connecting via ssh.

TODO: Use realtimeclock

Analise images instantly, in accordance to circumvent legal issues.

About

Using Computer Vision to Count Pedestrians. Hardware Documentation


Languages

Language:Shell 100.0%