Segmentation Fault in ArUco Detection on Raspberry Pi (ARM) with OpenCV 4.6.0

[jbneto1]

Environment

• Hardware: Raspberry Pi 5
• Operating System: Raspberry Pi OS (Bookworm)
• OpenCV Version: 4.6.0
• Python Version: 3.11
• Camera: PiCamera2 with libcamera v0.5.0, libpisp version v1.2.1

Description

When running ArUco marker detection on Raspberry Pi using the OpenCV 4.6.0 API methods, the program crashes with a segmentation fault. The crash happens specifically during the cv2.aruco.detectMarkers() call.

Steps to Reproduce

1. Set up a Python script that uses OpenCV's ArUco detection
2. Use the ArUco's: cv2.aruco.getPredefinedDictionary() and cv2.aruco.DetectorParameters()
3. Call cv2.aruco.detectMarkers() with the created dictionary and parameters
4. The program crashes with SIGSEGV at a very low memory address (0x60)

Crash Information

When run with strace, the program shows the following crash pattern:

[DEBUG timestamp] Memory usage before detection: 174.37109375 MB
--- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0x60} ---
+++ killed by SIGSEGV +++

Workaround

The issue can be resolved by using the older ArUco API functions:

• Use cv2.aruco.Dictionary_get() instead of cv2.aruco.getPredefinedDictionary()
• Use cv2.aruco.DetectorParameters_create() instead of cv2.aruco.DetectorParameters()

Code Sample to Reproduce

import cv2
import numpy as np

# Setup camera and capture frame
# ...

# This crashes on ARM/Raspberry Pi
aruco_dict = cv2.aruco.getPredefinedDictionary(cv2.aruco.DICT_5X5_50)
parameters = cv2.aruco.DetectorParameters()
corners, ids, rejected = cv2.aruco.detectMarkers(gray_image, aruco_dict, parameters=parameters)

# This works (using older API)
dictionary = cv2.aruco.Dictionary_get(cv2.aruco.DICT_5X5_50)
parameters = cv2.aruco.DetectorParameters_create()
corners, ids, rejected = cv2.aruco.detectMarkers(gray_image, dictionary, parameters=parameters)