This research addressed the challenges faced by small-scale egg producers in assessing egg quality through the development of a lightweight network that utilized the MobileNetV4 backbone for efficient lightweight feature extraction. In this research, we proposed the ODGC2f module and the OmniSPPF module for multi-scale feature attention. The model was evaluated on a dataset of 806 egg images collected from two farms, which categorized four distinct quality indicators: Broken Yolk, Chamber, Embryo, and Spoiled Area. Our proposed architecture reduced model parameters by 25.6% and computational complexity by 14.4% compared to the baseline model, while it maintained higher robust performance with 88.6% mAP@50 and 54.7% mAP@50:95 for object detection. This work demonstrated a practical solution for small-scale egg producers, as it enabled efficient and accurate quality assessment without requiring substantial computational resources.