Training a Face Detection Model on WIDER FACE Dataset

[AcSully]

I would like to train a face detection model using the WIDER FACE dataset.
I have already created the environment with Conda, and the training process ran normally. The training results are as follows:

Using object_detection/stm32ai_main.py:

python /home/niuke/STM32MP/Github/stm32ai-modelzoo-services/object_detection/stm32ai_main.py

Are these results normal? If not, what should I change?
Could you share an example of training a face detection model using the STM32 Model Zoo Services?

I am using the user_config.yaml file and the UART log file (attached).

user_config.yaml
uart_log.log

The prepared WIDER FACE dataset can be found here:

https://drive.google.com/file/d/1no40Mn4HTMX0Kvf0kpk8aA6IK-dIYhB8/view?usp=drive_link