Strabismus is an eye misalignment. It is a major cause of ir- reversible loss of vision called amblyopia. In Africa, where healthcare resources are frequently limited, countries such as Senegal encounter specific difficulties in the detection and management of strabismus as a result of these constraints. Timely detection of strabismus is essential to prevent the development of amblyopia. This paper evaluates the performance of several deep learning models in identifying strabismus early from facial images. Three classical deep learning models were used: MobileNetV2, ResNet-50, and CNNModel. Models performances were evaluated and compared, revealing that MobileNetV2 and ResNet-50 demonstrated comparable results on the test dataset. These findings highlight the strong potential of deep learning models to improve strabismus screening, especially in low-resource settings like Senegal, by increasing the accessibility and reliability of diagnosis.