This study investigates the impact of pre-image processing optimization to enhance detection, accuracy, brightness, and contrast adjustments on the performance of an underwater fish classification model. Using a dataset of 219 goldfish photos, five brightness and contrast adjustment levels are applied, ranging from minimal to maximal intensity (0–40). These adjustments resulted in 25 test cases (5 × 5 combinations), which were trained using the YOLOv11 model. The performance indicators assessed included accuracy, precision, recall, and F1 score. The results show that the best settings for accuracy and precision were adjusted to 30 for brightness and 30 for contrast. However, the F1 score was compared to evaluate the overall efficacy of the model. The optimal F1 score is 0.864, at 30 for contrast and 30 for brightness. The detection performance was compared with the baseline to validate the test results. It was found that the detection performance of the baseline was 70.58%, while the optimum point achieved 82.35%, respectively. The results indicate that effective brightness and contrast adjustments enhance object detection, especially in images with low-contrast backgrounds. The relationship between these adjustments and model performance was also nonlinear, emphasizing the need for hyperparameter tuning techniques to improve model accuracy and reliability under various contrast and brightness settings. Focusing on these adjustments can boost the model's effectiveness and adaptability to different image qualities. Moreover, these results can be applied to improve underwater fish detection models, ensuring optimal performance in varying environmental conditions.