This study evaluated the effectiveness of portable, low-cost techniques, specifically colorimetry and visible spectroscopy for discriminating dark, firm, and dry (DFD) beef carcasses during processing in a certified slaughterhouse. Portable, low-cost devices were used for real-time assessments. A total of 523 chilled carcasses were analyzed, of which 449 were classified as normal and 74 as DFD, based on pH measurements (≥5.8 for DFD). Colorimetric variables (L*, a*, b*, chroma, and hue) were consistently higher in normal carcasses compared to DFD. Supervised classification models were used to predict DFD status using the evaluated detection techniques. Given the class imbalance, the Synthetic Minority Over-sampling Technique (SMOTE) was applied, significantly improving the performance of machine learning models. Portable spectroscopy showed superior performance, especially when combined with preprocessing techniques such as multiplicative scatter correction and second-order derivatives, achieving 96.77 % sensitivity and 98.06 % specificity with the Random Forest model. Although colorimetry proved effective, spectroscopy yielded greater reliability for real-time DFD detection. These findings highlight the potential of these techniques as cost-effective alternatives to traditional methods, supporting their applicability in rapid and objective meat quality assessments under industrial conditions.