Inaccurate Slim Models

[gautamHCSCV]

I am using the weights of slim model as:

ocr = PaddleOCR(lang="en", show_log = True, use_gpu = False,
rec_model_dir = 'en_number_mobile_v2.0_rec_slim_infer/', det_model_dir ='en_PP-OCRv3_det_slim_infer',
)
result = ocr.ocr(IMAGE_PATH)

The result of using this model on a sample image is completely absurd. Am I making any mistake?
Could you provide a sample code of how to use the slim models for english text recognition on a sample image.