This study investigates whether retinotopic mapping can be integrated into classical computational architectures that model biological visual pathways, specifically deep convolutional neural networks (dCNNs), to enhance image categorization and localization performance. The study retrains a retinotopic dCNN on the ImageNet task, which involves classifying images into 1000 labels.The study's primary finding is that standard deep CNNs are highly effective in processing retinotopic inputs, despite the significant transformation of visual inputs that this mapping entails (A). The ResNet101 network is particularly adept at adapting to inputs where a large portion of the image is compressed in the periphery and the spatial arrangement is disrupted. Furthermore, we demonstrate that it enhances robustness to arbitrary image rotations, particularly for isolated objects (A) & (B). However, this invariance is achieved at the expense of reduced invariance to translations.Conversely, its ability to localize visual targets is improved due to its increased sensitivity to translations (C). Once trained, the network can detect objects by systematically shifting the center of gaze across the image (Table 1). This provides an object localization mechanism as a result of retinotopic mapping.These results support the interpretation of retinotopic mapping as a key component of object-driven preattentive visual processes. They also explain how eye movements interact synergistically with this architecture to always place objects of interest at the center of gaze.