Outsourcing the paranoid neighbor to machines
One machine (let's name him Jeff) sits at the window and takes pictures. If something happens Jeff tells another machine (Let's call her Lisa). Lisa takes a look at the picture and tells everyone about what they saw.
A Raspberry Pi with a webcam uploads photos to a webserver - if there is enough change to the previous one. The webserver analyses the picture with Google's Tensorflow and tweets the best match.
- Tweet sentences: Provide a dictionary with some randomly picked sentence: e.g. 'Just saw a {{word}} passing by.' or 'Oh look, a {{word}}!'
- AllInOne: Due to hardware limitiations it's currently not possible to install tensorflow on a raspberry - See this issue for details.
- Raspberry Pi or similar device
- cheap USB-webcam
- a webserver (Let's name her Lisa) for processing
cd <<project_source>>
openssl rand -base64 64 > .vault_passphrase
Create the vault - delete the existing file if necessary
ansible-vault create group_vars/jeff/vault.yml
Add this content to group_vars/jeff/vault.yml
:
---
vault_wifi_name:
vault_wifi_passphrase:
vault_lisa_token:
- Install the default raspbian-image
- Find out the raspberry's ip address
- I strongly recommend to enable passwordless SSH access on the raspberry and disable password-based login:
sudo vi /etc/ssh/sshd_config
and setPasswordAuthentication no
- Set the found ip address as the ip address of jeff in the file named
hosts
. - If you want to enable wifi, set the name and passphrase in
ansible-vault edit group_vars/jeff/vault.yml
ansible-playbook jeff.yml