Various embodiments of the present disclosure are directed towards methods for forming an image sensor in which a device layer overlies and has a different semiconductor material than a substrate and in which the device layer has high crystalline quality. Some embodiments of the methods include: epitaxially growing the device layer on the substrate; patterning the device layer to form a trench dividing the device layer into mesa structures corresponding to pixels; forming an inter-pixel dielectric layer filling the trench and separating the mesa structures; and forming photodetectors in the mesa structures. Other embodiments of the methods include: depositing the inter-pixel dielectric layer over the substrate; patterning the inter-pixel dielectric layer to form cavities corresponding to the pixels; epitaxially growing the mesa structures in the cavities; and forming the photodetectors in the mesa structures.