2019-10-10

motion project https://motion-project.github.io/

Auflösung der cam: https://picamera.readthedocs.io/en/release-1.12/fov.html

Als user motion die sqlite3 database erzeugen und tabelle anlegen

CREATE TABLE security (camera int, filename char(80) not null, frame int, file_type int, time_stamp timestamp without time zone, event_time_stamp timestamp without time zone);

Zu motion:

Motion hat das konzept von einem "(motion) event".
Ein event kann mehrere frames beinhalten
Wenn eine weitere motion innerhalb von event_gap sekunden erkannt wird, werden die frames zu dem gleichen event gezählt.
Events werden in der Datenbank über event_time_stamp identifiziert.
Mit post_capture kann man die anzahl der frames nach einer bewegung einstellen. wahrscheinlich wichtig für uns wegen motion blur
Wichtig threshold ist wahrscheinlich zu groß um bewegung auf einer kleinen bildflaeche zu erkennen

bei gleicher belichtung - maessiges zimmerlicht, nachmittags

3280x2464 -> ungefaehr 1fps
1920x1080 -> 3fps

Google Drive Folder für Daten https://drive.google.com/drive/folders/1CwqrDiGBhyMPjHcpFzJEgv0bM2oV608s?usp=sharing

rclone auf raspi pi konfiguriert.

Gelernt wie man cron jobs in systemd konfiguriert:

/home/pi/.local/share/systemd/user/foobar.service:

[Unit]
Description = some quux

[Service]
Type = oneshot
ExecStart = /home/pi/insects/bin/bla.sh
StandardOutput = journal

[Install]
WantedBy = default.target

/home/pi/.local/share/systemd/user/foobar.timer:

[Timer]
OnCalendar=*-*-* *:*:*

[Install]
WantedBy=default.target

OpenCV on raspberry

Install libaries

sudo apt-get install libatlas-base-dev sudo apt-get install libjasper-dev sudo apt-get install libqtgui4 sudo apt-get install python3-pyqt5 sudo apt-get install libqt4-test sudo apt-get install libilmbase-dev libopenexr-dev libgstreamer1.0-dev

2019-10-18

Reconfigure wifi:

edit /etc/wpa_supplicant/wpa_supplicant.conf
Reload with wpa_cli -i wlan0 reconfigure. (systemctl netorking doesnt seem to work)

You can also copy to /boot/wpa_supplicant.conf and restart if you have access to SD card.

76cm Breite, 42cm bei 61cm Abstand

Länge: 80cm-100cm Breite: 80cm Höhe: 50cm

open source ecology germany

-> 100x60x60

Get live picture via:

python3 ./tools/lastpic.py http://195.201.97.57:5000/ /media/usb/cam1/frames.db

Mount Server

Unmount

diskutil unmount force /Users/levin/volumens/eco1

Mount

sshfs -o IdentityFile=~/.ssh/id_eco_tunnel tunneldigger@195.201.97.57: /Users/levin/volumens/eco1

how to sync

gsutil copy -r /media/usb/cam1/frames gs://eco1/cam1/frames

Django Backend

Setup Database

python manage.py migrate

Load test data

python manage.py test

Start server

python ./manage.py runserver

Export Schema

./manage.py graphql_schema --out frontend/schema.json

Example Query

query { allFrames(url: "test1") { edges { node { id, url } } } }

query { allCollections { edges { node { frames { edges { node { id url } } } } } } }

Verbleibende Workpackages

Data Browser
- Daten-Modell
- React View - Browsing
- React View - Browsing verkabeln (Backend schreiben)
- React View - Warenkorb
- React View - Warenkorb verkabeln (Backend schreiben)
Annotation / Model Vizualazier
- React View
- React View verkabeln
Live View
Dataset-View
- verkabeln
Daten labelling
- Micrsoft- React component einbauen
- verkabeln
Data Scientist Toolkit
- Backend für read/write
- Python Library schreiben
Graph View von labelling über Zeit
- view
- verkabeln (backend schreiben)

Query Bla {
    frames(begin: DateTime!, end: DateTime!, nSample: Int) {
        id
        timestamp
        url
        thumbnailUrl
    }

    collections() {
    }

    collection(collectionId: String) {
        frames(begin: DateTime!, end: DateTime!) {
            labels: {
                processId
            }
        },
        labelProcesses : {
            id
            name
            dateCreated
        }

    }


}

2020-01-15

heimchen: acheta domesticus

Wed Jan 15 11:09:11 CET 2020 wanderheuschrecken (locasta migratoria)

Wed Jan 15 12:32:08 CET 2020 wuestenheuschrecken (schistocerca gregaria)

To start an ipython shell inside the django project:

APPENV=prod python3 manage.py shell

get a equidistant subsample

{
    frames(tbegin: "2019-11-01T00:00:00",
           tend: "2019-11-14T00:00:00",
           nframes: 10) {
      id,
      url,
      timestamp,
      thumbnail
    }
}

get a continuous page

{
    frames(tbegin: "2019-11-01T00:00:00",
           tend: "2019-11-15T00:00:00",
           after: "RnJhbWU6Nzk5NzU=",
           nframes: 10) {
      id,
      url,
      timestamp,
      thumbnail
    }
}

stack of: [b, e] set from outside: clear stack zoom in button: - if zoomable (select and domain not equal): - add domain to stack - set domain to [b, e] & selection to [b, e] zoom out button: - if stack not empty: - pop domain of stack - (adapt domain and selction so that selection is conatined in domain) - if stack is empty: - widen domain by x% in both directions

https://codesandbox.io/s/rw97j317p https://observablehq.com/@d3/d3-scaletime https://observablehq.com/@d3/scale-ticks

2019-11-15 Heimchen 2020-01-15 Wanderheuschrecken 2020-01-16 Wuestenheuschrecken

2020-02-15

Data Browser

right side bar (toggleable): datasets

bottom: create new data set (empty)
delete dataset / rename dataset
download link
"all" oder collection auswählen -> bestimmt kontext in dem man sucht

search

continuous / subsample view
tag filter

search results

zoom by ratio
evenly spaced
show results in view
add to collection -> to which one?, subsample / continuous?

Frame

multiple labels
a label has a "level" in the taxonomy (species, family, etc.)
bounding boxes editable
prev / next buttons only when context is collection

smatting/insects