| license: mit | |
| language: | |
| - en | |
| pipeline_tag: depth-estimation | |
| tags: | |
| - coreml | |
| Based on https://github.com/fabiotosi92/ZipDepth - Converted to CoreML for use on iPhone Devices, Internal testing suggested iPhone 12> will attain good results | |
| 📊 Quantitative Results | |
| ZipDepth achieves state-of-the-art accuracy among lightweight embedded models on NYUv2, KITTI, ETH3D, ScanNet, and DIODE, while being significantly more efficient than large pretrained models. | |
|  | |
| 🏗️ Architecture | |
|  | |
| The encoder is organized in four hierarchical stages. Stages 1–2 use RepVGG reparameterizable blocks (3×3 + 1×1 + identity branches fused into a single 3×3 at inference) augmented with Strip Pooling Attention for horizontal/vertical context. Stage 3 adds Squeeze-and-Excitation channel attention and a Global Context Block. Stage 4 deepens the representation with additional RepVGG blocks. | |
| The neck combines SPPF multi-scale pooling with a Cross-Scale Fusion module. The decoder is a lightweight FPN with a Convex Upsampling head for sub-pixel-accurate depth maps. | |