Detect.Confidence values

As mentioned in this post:

When detecting a cell phone Detect.Confidence shows 7.50.8399, could you explain this value?

Hi @james0.0joseph, if you are using jetson_inference.detectNet, that doesn’t look like the normal GUI or visualization, so I’m not sure at what point the confidence scores are going wrong. First, try printing the detections like this to see if they hold valid confidence values:

detections = net.Detect(img)

In my python code I do
detections=net.Detect(frame, width, height)
for detect in detections:
cv2.putText(img,str(round(fpsFilt,1))+ str(detect.Confidence) ,(y,x),font,1,(0,0,255),2)

Is there a way of extracting the confidence value like this or does it need to be done differently?

@james0.0joseph I would try formatting your confidence string like this and printing it to the terminal too, to confirm that it’s working for you as expected:

print(f"Confidence:  {detect.Confidence * 100.0:.1f}")

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.