HardFault_Handler issue when implementing st_yolo_x_nano on STM32N6570-DK

[Monk Eastman]

Hello,

At my workplace, we have purchased two STM32N6570-DK boards to implement our computer vision models. We are currently trying to implement the "st_yolo_x_nano 480x480x3" model from the ModelZoo.

We followed these two links, strictly respecting the versions indicated for both STM32CubeIDE and STEdge:

• How to run AI models from Model Zoo on STM32N6

• Introduction to Edge AI on STM32N6 (This second link uses STEdge 2.2, as per the reference).

The Issue: Every time I attempt to debug, the same thing happens: the program enters a HardFault_Handler(). This occurs due to an invalid/null pointer error returned by the function LL_ATON_Output_Buffers_Info() in buffers_info[0].scale.

The error happens specifically when it reaches line 36 of the code. According to the debugger, it cannot access memory at address 0x0.

Technical Details:

• Configuration: The app_Config.h file being used is the one provided in the ModelZoo.

 /**
 ******************************************************************************
 * @file    app_config.h
 * @author  GPM Application Team
 *
 ******************************************************************************
 * @attention
 *
 * Copyright (c) 2023 STMicroelectronics.
 * All rights reserved.
 *
 * This software is licensed under terms that can be found in the LICENSE file
 * in the root directory of this software component.
 * If no LICENSE file comes with this software, it is provided AS-IS.
 *
 ******************************************************************************
 */

#ifndef APP_CONFIG
#define APP_CONFIG

#include "arm_math.h"

#define USE_DCACHE

/*Defines: CMW_MIRRORFLIP_NONE; CMW_MIRRORFLIP_FLIP; CMW_MIRRORFLIP_MIRROR; CMW_MIRRORFLIP_FLIP_MIRROR;*/
#define CAMERA_FLIP CMW_MIRRORFLIP_NONE

#define ASPECT_RATIO_CROP       (1) /* Crop both pipes to nn input aspect ratio; Original aspect ratio kept */
#define ASPECT_RATIO_FIT        (2) /* Resize both pipe to NN input aspect ratio; Original aspect ratio not kept */
#define ASPECT_RATIO_FULLSCREEN (3) /* Resize camera image to NN input size and display a maximized image. See Doc/Build-Options.md#aspect-ratio-mode */
#define ASPECT_RATIO_MODE ASPECT_RATIO_CROP

/* Model Related Info */
#define POSTPROCESS_TYPE    POSTPROCESS_OD_ST_YOLOX_UI

#define NN_WIDTH      (480)
#define NN_HEIGHT     (480)
#define NN_BPP 3

#define COLOR_BGR (0)
#define COLOR_RGB (1)
#define COLOR_MODE    COLOR_RGB

/* Classes */
#define NB_CLASSES   (1)
#define CLASSES_TABLE const char* classes_table[NB_CLASSES] = {\
   "person"}\

/* I/O configuration */
/* Postprocessing ST_YOLO_X configuration */
#define AI_OD_ST_YOLOX_PP_NB_CLASSES        (1)
#define AI_OD_ST_YOLOX_PP_L_GRID_WIDTH      (60)
#define AI_OD_ST_YOLOX_PP_L_GRID_HEIGHT     (60)
#define AI_OD_ST_YOLOX_PP_L_NB_INPUT_BOXES  (AI_OD_ST_YOLOX_PP_L_GRID_WIDTH * AI_OD_ST_YOLOX_PP_L_GRID_HEIGHT)
#define AI_OD_ST_YOLOX_PP_M_GRID_WIDTH      (30)
#define AI_OD_ST_YOLOX_PP_M_GRID_HEIGHT     (30)
#define AI_OD_ST_YOLOX_PP_M_NB_INPUT_BOXES  (AI_OD_ST_YOLOX_PP_M_GRID_WIDTH * AI_OD_ST_YOLOX_PP_M_GRID_HEIGHT)
#define AI_OD_ST_YOLOX_PP_S_GRID_WIDTH      (15)
#define AI_OD_ST_YOLOX_PP_S_GRID_HEIGHT     (15)
#define AI_OD_ST_YOLOX_PP_S_NB_INPUT_BOXES  (AI_OD_ST_YOLOX_PP_S_GRID_WIDTH * AI_OD_ST_YOLOX_PP_S_GRID_HEIGHT)
#define AI_OD_ST_YOLOX_PP_NB_ANCHORS        (3)
static const float32_t AI_OD_ST_YOLOX_PP_L_ANCHORS[2*AI_OD_ST_YOLOX_PP_NB_ANCHORS] = {30.000000, 30.000000, 4.200000, 15.000000, 13.800000, 41.999999};
static const float32_t AI_OD_ST_YOLOX_PP_M_ANCHORS[2*AI_OD_ST_YOLOX_PP_NB_ANCHORS] = {15.000000, 15.000000, 2.100000, 7.500000, 6.900000, 21.000000};
static const float32_t AI_OD_ST_YOLOX_PP_S_ANCHORS[2*AI_OD_ST_YOLOX_PP_NB_ANCHORS] = {7.500000, 7.500000, 1.050000, 3.750000, 3.450000, 10.500000};
#define AI_OD_ST_YOLOX_PP_IOU_THRESHOLD      (0.5)
#define AI_OD_ST_YOLOX_PP_CONF_THRESHOLD     (0.6)
#define AI_OD_ST_YOLOX_PP_MAX_BOXES_LIMIT    (100)

/* Display */
#define WELCOME_MSG_1         "st_yolo_x_nano_480_1.0_0.25_3_st_int8.tflite"
#define WELCOME_MSG_2         "Model Running in STM32 MCU internal memory"

#endif

• Hardware Setup: The STM32N6570-DK is connected via USB-C (PD) to my laptop.

    What could be causing this?

    Thank you very much in advance for your help.

[Andrew Neil]

debugging Cortex-M Hard Faults

[Monk Eastman]

HI @Andrew Neil 

I edited the question to be more specific, and tried to follow the forum guidelines.

Thanks

[Julian E.]

Hi @Monk Eastman,

Do you use the st model zoo services scripts for deployment or do you do it manually?

What version of the ST Edge AI Core do you use? 2.2?

What do you observe on the board?

We are investigating an issue with yoloX models where, at reset, sometime the camera display works, sometimes we get a black screen and most importantly, after a random number of inferences, the bounding boxes freeze.

It seems to be because of a bug in the postprocessing. I am waiting for an answer from the dev team.

Have a good

Julian

[Andrew Neil]

I was linking specifically to the bit about how to debug Hard Faults on Cortex-M

[Julian E.]

@Monk Eastman,

On your image we see a DataType_Float; so, it is normal that scale and offset are NULL.

I think you need to use POSTPROCESS_OD_ST_YOLOX_UF (float) and not POSTPROCESS_OD_ST_YOLOX_UI (int)

I would suggest that you use the model zoo script deployment operation mode to deploy this model. It does a generate with options to have int8 output and manage automatically the app_config.h config. (Using the ui in this case of course).

Also, we fix the bug I mentioned in my last message. It was a postprocessing bug. So, if you encounter what I described, please wait for the model zoo script update 4.0, planned for end of January.

Have a good day,

Julian