A basic pipeline¶

This example shows a basic pipeline that reads some frames of a video and writes them in an output video.

[1]:

%load_ext autoreload
%autoreload 2

For this example we will create a pipeline with only two components: - A source that reads from a video file. - A sink that writes to a video file.

[2]:

from videoanalytics.pipeline import Pipeline
from videoanalytics.pipeline.sources import VideoReader
from videoanalytics.pipeline.sinks import VideoWriter

Note: the video used in this example was downloaded from youtube.

[3]:

DATA_PATH = "../data"

# Input
INPUT_VIDEO = DATA_PATH+"/input/test_video.mp4"
START_FRAME = 0
MAX_FRAMES = 100

[4]:

%%HTML
<div style="text-align: center">
    <video width="600" height="400" controls>
      <source src="../data/input/test_video.mp4" type="video/mp4">
    </video>
</div>

Path for output video.

[5]:

OUTPUT_VIDEO = DATA_PATH+ "/output/test_output.avi"

Typical workflow¶

The following steps illustrate the typical workflow for definining and executing a pipeline.

Step 1. Create the global context used by components to share data.

[6]:

context = {}

Step 2. Create the pipeline and add each component.

[7]:

pipeline = Pipeline()

[8]:

pipeline.add_component( VideoReader( "input",context,
                 video_path=INPUT_VIDEO,
                 start_frame=START_FRAME,
                 max_frames=MAX_FRAMES))

[9]:

pipeline.add_component(VideoWriter("writer",context,filename=OUTPUT_VIDEO))

Step 3. Define the order of execution. Each tuple (A,B) indicates that B has to be executed after A.

[10]:

pipeline.set_connections([
    ("input", "writer")
])

Step 4 (optional). Remove isolated components and plot the graph.

[11]:

pipeline.optimize()

[12]:

import matplotlib.pyplot as plt

fig,axes = plt.subplots(1,1,figsize=(22,8))
pipeline.plot(ax=axes)

Step 5. Execute the pipeline.

[13]:

pipeline.execute()
print("Total execution time [s]:", pipeline.get_total_execution_time())

Total execution time [s]: 2.886986176999926

Step 6 (optional). Report the execution times for each component. Times are expressed in seconds.

[14]:

import pandas as pd

metrics_df = pd.DataFrame.from_dict(pipeline.get_metrics(), orient='index',columns=["time"])
metrics_df

[14]:

	time
input_avg_dt	0.006297
writer_avg_dt	0.022153

Display the output video.

Note: currently XVID format is not supported by jupyter.

[15]:

%%HTML
<div style="text-align: center">
    <video width="600" height="400" controls>
      <source src="../data/output/test_output.avi" type="video/avi">
    </video>
</div>