This research delves into Camouflage Object Detection (COD), with a primary objective of accurately identifying and segmenting objects that adeptly blend into their surroundings. Our innovation lies in refining the COD model's loss function by breaking it down into three pivotal constituents: Binary Cross-Entropy Loss, IoU Loss, and Dice Loss. These components are meticulously weighted and then linearly combined to formulate the wBID Loss. Our proposed model, wBID-SNet, synergistically incorporates this advancement with the SARNet model, undergoing rigorous training, validation, and testing phases across the CAMO, COD10K, and NC4K datasets.Through comprehensive experimentation, our methodology demonstrates a substantial boost in performance compared to existing benchmarks. Across the diverse datasets, our approach yields consistent improvements, showcasing average enhancements of 0.579%, 2.512%, and 1.473% in S, F, and E metrics, respectively, when juxtaposed with FSPNet. Furthermore, on the NC4K dataset, our approach exhibits even more pronounced advancements, with increases of 1.362%, 3.255%, and 1.656%, respectively. These findings underscore the efficacy of our proposed methodology in significantly augmenting the accuracy and efficacy of camouflage object detection systems.