With the complex composition of the radioisotopes and waste materials, the characterization of the volumetric low-level wastes from CANDU plants is challenging. This study presents a technique to localize and quantify the contaminations presented in the CANDU waste containers. MCNP-based models are developed for an N-type coaxial HPGe detector and a LaBr3 detector to simulate the photon peak information. The simulated efficiency and the experimental count rates are combined to estimate the activity of unknown waste samples. During the spectrum collection of a 4L Marinelli beaker source and 1-quart waste samples, the MCNP algorithm showed better accuracy in activity estimation than the Mirion ISOCS/LabSOCS software. With further development, this method has the potential to outperform the popular commercial software in estimating activity for volume sources with complex geometry and uneven distribution. The multi-detector array models with hotspot designs are also studied in this work to provide real-time information about the location and activity of the contamination inside the 2.2 m3 industrial low-level waste containers. The on-site measurements show promising results as the position of the contamination was able to be located within a volume of 61×40×34 cm. Overall, this technique has good potential to be utilized in the nuclear industry for large-volume low-level waste analysis. / Thesis / Master of Science (MSc)