Is there are way to do classification for Video

#1

I trained a model that can detect few objects. I have a video(locally saved in my computer) that needs to test with that model.Is there a way to do that?

#2

@dilanperera There’s an example in the Python SDK that classifies data per frame here https://github.com/edgeimpulse/linux-sdk-python/blob/master/examples/image/classify-image.py

Then you can extract the frames one by one with OpenCV, see e.g. https://stackoverflow.com/questions/33311153/python-extracting-and-saving-video-frames and classify them.

Suggested development box to run & debug edge-impulse-linux and OpenCV
#3

Will try…Thank you very much

#4

Is it only for a Linux? how can I do it on a windows machine?

#5

Correct, we don’t support Windows. You can:

  1. Get the TensorFlow Lite (int8 quantized) model from Dashboard (under ‘Download block output’).
  2. Install TensorFlow Lite on Windows (https://www.tensorflow.org/lite/guide/python)
  3. Pass an image from OpenCV directly into TensorFlow from Python (https://heartbeat.comet.ml/running-tensorflow-lite-image-classification-models-in-python-92ef44b4cd47, steps 3 and further)

I think that would work on Windows, but outside the scope where we can do much to support you.

2 Likes
Edge Impulse video input source
#6

It worked…Thank you… Awsome Platform…

1 Like