bioroboticslab / beesbook_backend Goto Github PK

Python 7.82% Jupyter Notebook 92.18%

beesbook_backend's Introduction

beesbook_backend

The purpose of this backend is to interface the data on the servers and the user and developers. It currently holds the ability to plot videos and images from bb_binary and bb_tracking data.. The plots can be cropped to show only relevant areas and the filling of gaps (for missing frames in tracks).

How To (for users)

install api part of bb_backend (no dependencies required)
- pip install git+https://github.com/BioroboticsLab/beesbook_backend.git
tunnel

ssh -N -L 8000:localhost:8000 thekla.imp.fu-berlin.de

use bb_backend.api to fetch images and videos:

import bb_backend.api
from bb_backend.api import FramePlotter, VideoPlotter
bb_backend.api.server_adress = '127.0.0.1:8000'

frame_plotter = FramePlotter(frame_id=12058089920919369671, xs=[350, 450], ys=[350, 450],
                                sizes=[40,200],title="Image title (e.g. timestamp)",
                                labels=["Abee","Beebee"], scale=1.0, colors=["r", "y"])
plt.imshow(frame_plotter.get_image())

# Make video consisting of 3 times the same frame.
video_plotter = VideoPlotter(frames=[frame_plotter, frame_plotter, frame_plotter])

video_plotter.get_video(display_in_notebook=True)

Setup (not for users)

Install software and library requirements:

docker
docker-compose (python package)
python virtualenv with packages in requirements.txt
ffmpeg with x264 encoding and h264 decoding capability

Initialize database data volume:

docker create -v /var/lib/postgresql --name postgres_data postgres:9.6

docker-compose up -d

docker-compose exec postgres psql -U postgres -c "CREATE DATABASE beesbook;"

python manage.py migrate

Find and store all video locations

python manage.py make_db_video [video_path_location]

e.g. python manage.py make_db_video /home/beesbook/hlrn/work1-hlrn/videos_HD_2016/

Read all repository data

python manage.py make_db_repo [repo_folder]

e.g. python manage.py make_db_repo /mnt/storage/beesbook/repo_season_2016_fixed

Run the server

Please configure nginx or something similar

For testing python manage.py runserver is fine.

beesbook_backend's People

Contributors

Watchers

Forkers

walachey

beesbook_backend's Issues

frontend library to abstract interface

Knowing how to use the backend is kinda complicated. Build some functions to abstract the interface so other people can start using it too.

See plot.py from the old code for integration into notebooks.

deploy with uwsgi + nginx for more stability

steps:

make production settingsfile
configure nginx and uwsgi

Documentation

A documentation is required for following things:

How to deploy the application
- docker and postgres
- make_db_repo and make_db_video
- ffmpeg requirements
How to use the api
- getting images and videos (syntax of the post request)
- plotting images and videos (syntax of the post request)

check if the requested frame is even in the video

e.g. frame 5000 when video only has 1024

option to keep aspect ratio when cropping

The cropping currently allows videos with very weird aspect ratios. Write a routine to pad the video to a specific aspect ratio

blocking ffmpeg commands

The ffmpeg commands should block other ffmpeg commands until they are finished and then check if the files exist.

This is required when two parallel requests are made which extract the same video.

crop and fillgap not working together

The frames filled in with fillgap are currently not cropped.

plot_frame should crop or highlight focused bee

implement option to crop a frame image[or video if you are awesome] to a bee

info: requirements.txt nutzen noch eine alte bb_binary version

@walachey In den requirements ist noch eine alte bb_binary version, mit einer neuen version wurden xpos und ypos vertauscht wenn ich mich richtig erinnere. Das kann zu problemen bei plotten führen.

move caching to redis

The memoizing decorator functions (filepath_cacher, cacher) should use redis instead a file system.

It also should expire some of the items (e.g. extracted images) every now and probably even delete from the hard disk.

encoding of new videos does not use gpu when flag is set

gpu flag set true does not enable gpu accelerated encoding:

ffmpeg_frames_to_video = ' '.join([
    f'ffmpeg -v {verbosity_level}',
    '-r 3',  # input framerate
    '-i {input_path}',
    '-r 3',  # video framerate (kinda irrelevant)
    '-pix_fmt yuv420p',
    '-vcodec libx264',
    '{output_path}'
])

Internal Server Error: /plotter/plot_video/
Traceback (most recent call last):
  File "/home/ktugan/miniconda3/envs/backend/lib/python3.6/site-packages/django/core/handlers/exception.py", line 39, in inner
    response = get_response(request)
  File "/home/ktugan/miniconda3/envs/backend/lib/python3.6/site-packages/django/core/handlers/base.py", line 187, in _get_response
    response = self.process_exception_by_middleware(e, request)
  File "/home/ktugan/miniconda3/envs/backend/lib/python3.6/site-packages/django/core/handlers/base.py", line 185, in _get_response
    response = wrapped_callback(request, *callback_args, **callback_kwargs)
  File "/home/ktugan/miniconda3/envs/backend/lib/python3.6/site-packages/django/views/decorators/csrf.py", line 58, in wrapped_view
    return view_func(*args, **kwargs)
  File "/home/ktugan/beesbook_backend/plotter/views.py", line 79, in plot_video
    path = Frame.plot_video(data, fill_gap, crop)
  File "/home/ktugan/beesbook_backend/plotter/utils.py", line 35, in check_file
    mem_object = f.call_and_shelve(*args, **kwargs)
  File "/home/ktugan/miniconda3/envs/backend/lib/python3.6/site-packages/joblib/memory.py", line 476, in call_and_shelve
    _, argument_hash, metadata = self._cached_call(args, kwargs)
  File "/home/ktugan/miniconda3/envs/backend/lib/python3.6/site-packages/joblib/memory.py", line 430, in _cached_call
    out, metadata = self.call(*args, **kwargs)
  File "/home/ktugan/miniconda3/envs/backend/lib/python3.6/site-packages/joblib/memory.py", line 675, in call
    output = self.func(*args, **kwargs)
  File "/home/ktugan/beesbook_backend/plotter/models.py", line 113, in plot_video
    return media.plot_video(data, crop)
  File "/home/ktugan/beesbook_backend/plotter/media.py", line 216, in plot_video
    xs = [x for p in data for x in p['x']]
  File "/home/ktugan/beesbook_backend/plotter/media.py", line 216, in <listcomp>
    xs = [x for p in data for x in p['x']]
KeyError: 'x'