--- library_name: pytorch license: other tags: - android pipeline_tag: image-to-image --- ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/ddcolor/web-assets/model_demo.png) # DDColor: Optimized for Qualcomm Devices DDColor is a coloring algorithm that produces natural, vivid color results from incoming black and white images. This is based on the implementation of DDColor found [here](https://github.com/piddnad/DDColor/). This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/ddcolor) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary). Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device. ## Getting Started There are two ways to deploy this model on your device: ### Option 1: Download Pre-Exported Models Below are pre-exported model assets ready for deployment. | Runtime | Precision | Chipset | SDK Versions | Download | |---|---|---|---|---| | ONNX | w8a16 | Universal | QAIRT 2.42, ONNX Runtime 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/ddcolor/releases/v0.50.2/ddcolor-onnx-w8a16.zip) | TFLITE | w8a8 | Universal | QAIRT 2.43, TFLite 2.19.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/ddcolor/releases/v0.50.2/ddcolor-tflite-w8a8.zip) For more device-specific assets and performance metrics, visit **[DDColor on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/ddcolor)**. ### Option 2: Export with Custom Configurations Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/ddcolor) Python library to compile and export the model with your own: - Custom weights (e.g., fine-tuned checkpoints) - Custom input shapes - Target device and runtime configurations This option is ideal if you need to customize the model beyond the default configuration provided here. See our repository for [DDColor on GitHub](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/ddcolor) for usage instructions. ## Model Details **Model Type:** Model_use_case.image_editing **Model Stats:** - Model checkpoint: ddcolor_paper_tiny.pth - Input resolution: 224x224 - Number of parameters: 56.3M - Model size (float): 215 MB - Model size (w8a8): 54.8 MB ## Performance Summary | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit |---|---|---|---|---|---|--- | DDColor | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 1773.576 ms | 19 - 4589 MB | NPU | DDColor | ONNX | w8a16 | Snapdragon® X2 Elite | 1798.539 ms | 165 - 165 MB | NPU | DDColor | ONNX | w8a16 | Snapdragon® X Elite | 3785.722 ms | 164 - 164 MB | NPU | DDColor | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 2957.561 ms | 37 - 5147 MB | NPU | DDColor | ONNX | w8a16 | Qualcomm® QCS6490 | 1809.322 ms | 307 - 313 MB | CPU | DDColor | ONNX | w8a16 | Qualcomm® QCS9075 | 5546.288 ms | 36 - 39 MB | NPU | DDColor | ONNX | w8a16 | Qualcomm® QCM6690 | 952.625 ms | 277 - 296 MB | CPU | DDColor | ONNX | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 2391.974 ms | 37 - 4671 MB | NPU | DDColor | ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | 908.979 ms | 421 - 441 MB | CPU | DDColor | ONNX | w8a8 | Snapdragon® X2 Elite | 1765.523 ms | 208 - 208 MB | NPU | DDColor | ONNX | w8a8 | Snapdragon® X Elite | 3145.372 ms | 207 - 207 MB | NPU | DDColor | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 936.536 ms | 0 - 542 MB | NPU | DDColor | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1210.022 ms | 0 - 616 MB | NPU | DDColor | TFLITE | w8a8 | Qualcomm® QCS6490 | 705.505 ms | 94 - 239 MB | CPU | DDColor | TFLITE | w8a8 | Qualcomm® QCS8275 (Proxy) | 3059.132 ms | 1 - 518 MB | NPU | DDColor | TFLITE | w8a8 | Qualcomm® QCS8550 (Proxy) | 1730.399 ms | 0 - 4 MB | NPU | DDColor | TFLITE | w8a8 | Qualcomm® SA8775P | 1729.61 ms | 0 - 451 MB | NPU | DDColor | TFLITE | w8a8 | Qualcomm® QCS9075 | 1600.906 ms | 0 - 61 MB | NPU | DDColor | TFLITE | w8a8 | Qualcomm® QCM6690 | 1682.182 ms | 9 - 354 MB | CPU | DDColor | TFLITE | w8a8 | Qualcomm® QCS8450 (Proxy) | 2113.256 ms | 0 - 662 MB | NPU | DDColor | TFLITE | w8a8 | Qualcomm® SA7255P | 3059.132 ms | 1 - 518 MB | NPU | DDColor | TFLITE | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 938.973 ms | 0 - 510 MB | NPU | DDColor | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 465.144 ms | 95 - 406 MB | CPU ## License * The license for the original implementation of DDColor can be found [here](https://github.com/piddnad/DDColor/blob/master/LICENSE). ## References * [DDColor: Towards Photo-Realistic Image Colorization via Dual Decoders](https://arxiv.org/abs/2201.03545) * [Source Model Implementation](https://github.com/piddnad/DDColor/) ## Community * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI. * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).