Hello,
Is it possible to run a tensorflow regression model on openmv?
I’ve created a regression model with edgeimpulse but when I run the inference on the MCU I only get the confidence value (as I have no labels).
Thanks
Hello,
Is it possible to run a tensorflow regression model on openmv?
I’ve created a regression model with edgeimpulse but when I run the inference on the MCU I only get the confidence value (as I have no labels).
Thanks
You can run a classification model. What do you mean by a regression model? What are you trying to do?
I have a mechanical crackmeter (see example ) and have trained a regression model with edge impulse to predict the x axis deplacement. The live classification gives me a scalar value on edgeimpulse platform but when I deploy on my vision Sheild
I 'm not able to get the scalar value, only the confidence score.
This is the same isn’t it though? The confidence score is a value between 0->1 or 0->255?
It s a value between 0 & 1
Can you post your code and the output?
the code :
import sensor, image, time, os, tf
sensor.reset() # Reset and initialize the sensor.
sensor.set_pixformat(sensor.GRAYSCALE) # Set pixel format to RGB565 (or GRAYSCALE)
sensor.set_framesize(sensor.QVGA) # Set frame size to QVGA (320x240)
sensor.set_windowing((240, 240)) # Set 240x240 window.
sensor.skip_frames(time=2000) # Let the camera adjust.
net = "trained.tflite"
clock = time.clock()
while(True):
clock.tick()
img = sensor.snapshot()
# default settings just do one detection... change them to search the image...
for obj in tf.classify(net, img, min_scale=1.0, scale_mul=0.8, x_overlap=0.5, y_overlap=0.5):
print("**********\nPredictions at [x=%d,y=%d,w=%d,h=%d]" % obj.rect())
img.draw_rectangle(obj.rect())
predictions =obj.output()
for pred in predictions:
print(pred)
print(clock.fps(), "fps")
as per the serial output :
Predictions at [x=0,y=0,w=240,h=240]
0.372549
6.19775 fps
**********
Predictions at [x=0,y=0,w=240,h=240]
0.341176
6.19779 fps
**********
Predictions at [x=0,y=0,w=240,h=240]
0.313725
6.19783 fps
**********
Predictions at [x=0,y=0,w=240,h=240]
0.262745
6.19786 fps
**********
Predictions at [x=0,y=0,w=240,h=240]
0.215686
6.19779 fps
**********
Predictions at [x=0,y=0,w=240,h=240]
0.235294
6.19783 fps
**********
Predictions at [x=0,y=0,w=240,h=240]
0.278431
6.19787 fps
**********
Predictions at [x=0,y=0,w=240,h=240]
0.184314
6.1979 fps
**********
Predictions at [x=0,y=0,w=240,h=240]
0.0705882
6.19794 fps
**********
Predictions at [x=0,y=0,w=240,h=240]
0.278431
6.19797 fps
**********
Predictions at [x=0,y=0,w=240,h=240]
0.376471
6.1979 fps
**********
Predictions at [x=0,y=0,w=240,h=240]
0.262745
6.19794 fps
**********
Predictions at [x=0,y=0,w=240,h=240]
0.490196
6.19798 fps
**********
thanks
Just multiply 0.313725 * 255
The 0.313725 is just the output you see in Edge Impulse divided by 255.