minimal python code to read out the raw data from an infiray p2pro camera and convert it to temperature

p2prolive.py

import cv2
import numpy as np

'''
with info from , check out:
https://www.eevblog.com/forum/thermal-imaging/infiray-and-their-p2-pro-discussion/200/
https://github.com/leswright1977/PyThermalCamera/blob/main/src/tc001v4.2.py

I did create a seperate conda env for this project on Windows:
conda create -n p2pro python
activate p2pro
pip install opencv-python pyusb pyaudio  ffmpeg-python

tested with cv2 version 4.8.0
'''

id = 0 # the p2pro camera may have a higher id (1,2..)
cap = cv2.VideoCapture(id) 
cap.set(cv2.CAP_PROP_CONVERT_RGB, 0) # do not create rgb data!

i = 0
while(True):    
    # Capture frame-by-frame    
    ret, frame = cap.read() 
    # frame is a 1D numpy array of 8bit integers, we need to reshape it
    # 1. index is the upper/lower image part of the video stream
    # the 4. index is to address the 2 bytes of the 16bit data
    frame = np.reshape(frame[0],(2,192,256,2))
    raw = frame[1,:,:,:].astype(np.intc) # select only the lower image part
    raw = (raw[:,:,1] << 8) + raw[:,:,0] # assemble the 16bit word
        
    temp = raw/64 - 273.2 # convert to Celsius scale
    
    i += 1
    if i%20 == 0 :
        print(f"min = {temp.min():1.4}, max = {temp.max():1.4}, avg = {temp.mean():1.4},")
    
    brightness = 0.01
    contrast = 0.95
     # values scaled to [0,1] range
    cv2.imshow('frame',(temp-temp.min())/(temp.max()-temp.min()) * contrast + brightness)
            
    #Waits for a user input to quit the application    
    if cv2.waitKey(1) & 0xFF == ord('q'):    
        break
            
cap.release()
cv2.destroyAllWindows()

I also made a version with a class definition that can be used in other projects directly:
https://gist.github.com/ks00x/9003fc0e1103bb2a4ecc690ab855633e

Line 28 is throwing ValueError: cannot reshape array of size 768 into shape (2,192,256,2).
Interestingly, frame.shape will print (384, 256, 3), so I presume the frame has been converted to RGB/BGR. Any idea what could cause this?

If the (in Windows required) cv2.CAP_DSHOW is omitted in line 18, the resulting array will have a shape of (1, 460800), which, to my mind, is not a multiple of either 256 and 192. Any help appreciated!

Line 28 is throwing ValueError: cannot reshape array of size 768 into shape (2,192,256,2). Interestingly, frame.shape will print (384, 256, 3), so I presume the frame has been converted to RGB/BGR. Any idea what could cause this?

If the (in Windows required) cv2.CAP_DSHOW is omitted in line 18, the resulting array will have a shape of (1, 460800), which, to my mind, is not a multiple of either 256 and 192. Any help appreciated!

could you please try: https://github.com/ks00x/p2proviewer/blob/main/p2profile.py which should work. If it doesn't, you might have modified the file in some way (edited metadata)