A camera class based of the popular VidGear video processing library. Neurosec seemlessly allows you to process inference using the Akida neuromorphic processor.
Project description
Neurosec
A camera class based off the popular VidGear video processing library. Neurosec seemlessly allows you to process inference using the Akida neuromorphic processor.
For best results, ensure you have an Akida neuromorphic processor installed. If you do not have one, you can purchase one from Brainchip Inc.:
How to install
Install via pip:
pip install neurosec
How to use Neurosec
Here is a simple example of using Neurosec
from neurosec import Neurosec
yolo_face = {
"fbz": "models/yolo_face.fbz",
"predict_classes": False,
"anchors": [[0.90751, 1.49967], [1.63565, 2.43559], [2.93423, 3.88108]],
"classes": 1,
"labels": {
0: "face",
},
"colours": {0: (255, 0, 0)},
"pred_conf_min": 0.70,
}
neurosec = Neurosec(
source=0,
model=yolo_face,
resolution=(640, 480),
).start()
while True:
frame = neurosec.get_neurosec_frame()
if frame is None:
break
cv2.imshow("Output", frame)
key = cv2.waitKey(1) & 0xFF
if key == ord("q"):
break
Since Neurosec leans entirely on VidGear - all of the amazing options are available, like streaming from youtube.
neurosec = Neurosec(
source="https://www.youtube.com/watch?v=crddAe9N2aM",
stream_mode=True,
model={
"fbz": "models/yolo.fbz",
"predict_classes": False,
"anchors": [
[0.56615, 1.05064],
[1.09098, 2.04053],
[2.384, 3.00597],
[2.45964, 4.91562],
[5.16724, 5.56961],
],
"classes": 2,
"labels": {
0: "car",
1: "person",
},
"colours": {0: (255, 255, 0), 1: (255, 0, 0)},
"pred_conf_min": 0.70,
},
).start()
A complete example:
from neurosec import Neurosec
if __name__ == "__main__":
try:
neurosec = Neurosec(
source=0,
model=MODELS["yolo_face"],
resolution=(640, 480),
).start()
while True:
frame = neurosec.get_neurosec_frame()
if frame is None:
break
cv2.imshow("Output", frame)
key = cv2.waitKey(1) & 0xFF
if key == ord("q"):
break
except KeyboardInterrupt:
print("die")
You can get frame meta at any time by calling:
meta = neurosec.get_frame_meta()
meta_json = get_frame_meta_json()
NeurosecNode
A simple Flask based web app that provides four main endpoints:
- {host} # generic view with streaming video embed
- {host}/feed/ # streaming video
- {host}/feed/overlay/ # strewaming video with overlay
- {host}/meta/ # frame meta
Accessing the server
An example request trying to access the nodes IP:
Through your browser:
Go to your computer or the devices IP address: http://10.0.0.1:5000?node_key={your_key}
Example meta output
import requests
your_node_key = "abcdefg"
node_ip_address = "http://10.0.0.2:5000/"
url = "{}meta".format(node_ip_address)
headers = {"X-Node-Key": your_node_key}
meta = requests.get(url, headers=headers)
print(meta.json())
or visit: http://10.0.0.1:5000/meta?node_key={your_key}
An example output while running Yolo:
{"timestamp": "1657216671.52087", "original_resolution": [640, 480], "decoded": [[231.37685351750486, 95.64780969570069, 434.0158447009765, 340.4589660097876, 0.0, 0.9620453119277954]], "labels": {"0": "face"}, "colours": {"0": [255, 0, 0]}, "pred_conf_min": 0.7}
How to run Neurosec-node
An example to start a server that will stream from the camera located at /dev/video0
from stdpio import NeurosecNode
neurosec_node = NeurosecNode(
**{
"source": 0,
"resolution": (640, 480),
"host": "0.0.0.0",
"node_key": "this_is_a_passw0rd",
"model": {
"fbz": "models/yolo_face.fbz",
"predict_classes": False,
"anchors": [
[0.90751, 1.49967],
[1.63565, 2.43559],
[2.93423, 3.88108],
],
"classes": 1,
"labels": {
0: "face",
},
"colours": {0: (255, 0, 0)},
"pred_conf_min": 0.70,
},
}
).run()
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distributions
Built Distribution
Hashes for neurosec-0.0.2-py2.py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 0e4428db0a605facdb54cb06a17018bf1dd65ba4a847a5b5f96a09d029c9b3b0 |
|
MD5 | f1fd0b4b07396e122374525f64820d9b |
|
BLAKE2b-256 | 17972ebbd1aef6ccbb6c9cf803b8ecd03a85f11464dcfa1c977ab6c870128a06 |