请提供下述完整信息以便快速定位问题/Please provide the following information to quickly locate the problem

• 系统环境/System Environment： Intel CPU Ubuntu 20.04
• 版本号/Version：Paddle： PaddleOCR： 问题相关组件/Related components： paddleOCR 2.5.0.3

First of all, thank you very much for developing such good software.

I am running PaddleOCR in CPU mode, and whenever I perform inference, there is a problem that the memory usage keeps increasing. Is there any way to limit memory usage?

Paddle OCR Initialization Code

paddle_ch = PaddleOCR(
            show_log=False,
            lang="ch",
            cpu_threads=1,
            det_db_thresh=0.1,
            det_db_box_thresh=0.1,
            use_mp=True,
            enable_mkldnn=True,
            total_process_num=os.cpu_count() * 2 - 1,
            use_angle_cls=True,
            cls_model_dir="whl/cls/ch_ppocr_mobile_v2.0_cls_infer",
            det_model_dir="whl/det/ch/ch_PP-OCRv3_det_infer",
            rec_model_dir="whl/rec/ch/ch_PP-OCRv3_rec_infer"
    )

Memory Profiler Result of PaddleOCR

ocr_1  | Line #    Mem usage    Increment  Occurrences   Line Contents
ocr_1  | =============================================================
ocr_1  |    108                                                         # 2. OCR
ocr_1  |    109    556.8 MiB      0.0 MiB          1                   try:
ocr_1  |    110    556.8 MiB      0.0 MiB           1                       if model.language in self.paddle_ocr:
ocr_1  |    111    938.6 MiB    381.8 MiB          1                           paddle_ocr_result = self.paddle_ocr[model.language].ocr(crop_img, cls=False)