edge-ml 0.3.3

Python library of edge-ml.org: end-to-end machine learning for embedded devices

Obtain data from edge-ml

from edgeml import DatasetReceiver

1. Create a project

This will also pull the metadata from the server. Here you need the read-key. Make sure there is not trailing / in your URL.

project = DatasetReceiver("https://edge-ml-beta.dmz.teco.edu", "8c051972b56e6b4ad6bd0bf573da580f")
print(project)

Dataset - Name: W_001, ID: 645a255bdd19d537f2a50126, Metadata: {}
Dataset - Name: W_004, ID: 645a255b7d9569d03843a12b, Metadata: {}
Dataset - Name: Square_003, ID: 645a255b2ce253c26b52426c, Metadata: {}
Dataset - Name: Square_004, ID: 645a255b1d6af5e7ed04575f, Metadata: {}
Dataset - Name: Square_002, ID: 645a255bc21261d4cbdc990d, Metadata: {}
Dataset - Name: W_002, ID: 645a255b3ebe0af02b211d5d, Metadata: {}
Dataset - Name: W_003, ID: 645a255b074737af782167e2, Metadata: {}
Dataset - Name: Square_001, ID: 645a255bbacf5ab8ff044304, Metadata: {}
Dataset - Name: edgemlDemo, ID: 64635a79a7a7513e8ac92d32, Metadata: {'langauge': 'python'}
Dataset - Name: edgemlDemo, ID: 64635a7b6fec1d2800b1cd06, Metadata: {'langauge': 'python'}

2. Actually obtain data

Until now, we only have metadata available. We need to pull the actual time-series data using one of the following methods:

We can load the data for a single timeSeries

project.datasets[0].timeSeries[0].loadData()
project.datasets[0].timeSeries[0].data.head()

<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; }

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	time	x
0	1970-01-01 00:10:23.339	-823.0
1	1970-01-01 00:10:23.378	-819.0
2	1970-01-01 00:10:23.418	-770.0
3	1970-01-01 00:10:23.458	-746.0
4	1970-01-01 00:10:23.497	-783.0

For a single dataset

project.datasets[0].loadData()
project.datasets[0].data.head()

<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; }

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	time	x	y	z	Gestures
0	1970-01-01 00:10:23.339	-823.0	-45.0	4025.0
1	1970-01-01 00:10:23.378	-819.0	-158.0	4075.0
2	1970-01-01 00:10:23.418	-770.0	-255.0	4116.0
3	1970-01-01 00:10:23.458	-746.0	-155.0	4059.0
4	1970-01-01 00:10:23.497	-783.0	-104.0	3963.0

Or for all datasets in the project

project.loadData()
project.data[0].head()

<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; }

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	time	x	y	z	Gestures
0	1970-01-01 00:10:23.339	-823.0	-45.0	4025.0
1	1970-01-01 00:10:23.378	-819.0	-158.0	4075.0
2	1970-01-01 00:10:23.418	-770.0	-255.0	4116.0
3	1970-01-01 00:10:23.458	-746.0	-155.0	4059.0
4	1970-01-01 00:10:23.497	-783.0	-104.0	3963.0

Upload data to edge-ml

1. Upload data using timestamps from the device

Here you will need the write-key

from edgeml.edgeml import DatasetCollector
sender = DatasetCollector(url="https://edge-ml-beta.dmz.teco.edu", apiKey="4e6159c9c77124d71f298e93f1ed7254", name="edgemlDemo", useDeviceTime=True, timeSeries=["accX", "accY"], metaData={"langauge": "python"})

import time

for i in range (100):
    await sender.addDataPoint(name="accX", value=i*0.1)
    await sender.addDataPoint(name="accY", value=i*0.5)
    time.sleep(0.01)
sender.onComplete()

True

2. Provide your own timestamps

from edgeml.edgeml import DatasetCollector
sender = DatasetCollector(url="https://edge-ml-beta.dmz.teco.edu", apiKey="4e6159c9c77124d71f298e93f1ed7254", name="edgemlDemo", useDeviceTime=False, timeSeries=["accX", "accY"], metaData={"langauge": "python"})

import time

for i in range (100):
    await sender.addDataPoint(timestamp=i*1000, name="accX", value=i*0.1)
    await sender.addDataPoint(timestamp=i*1000, name="accY", value=i*0.5)
    time.sleep(0.01)
sender.onComplete()

True