--- library_name: pytorch license: other tags: - real_time - android pipeline_tag: object-detection --- ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/deepbox/web-assets/model_demo.png) # 3D-Deep-BOX: Optimized for Mobile Deployment ## Real-time 3D object detection 3D Deep Box is a machine learning model that predicts 3D bounding boxes and classes of objects in an image. This model is an implementation of 3D-Deep-BOX found [here](https://github.com/skhadem/3D-BoundingBox/). This repository provides scripts to run 3D-Deep-BOX on Qualcomm® devices. More details on model performance across various devices, can be found [here](https://aihub.qualcomm.com/models/deepbox). ### Model Details - **Model Type:** Model_use_case.object_detection - **Model Stats:** - Model checkpoint: YOLOv3-tiny - Input resolution(YOLO): 224x640 - Input resolution(VGG): 224x224 - Number of parameters (Yolo2DDetection): 9.78M - Model size (Yolo2DDetection) (float): 37.3 MB - Number of parameters (VGG3DDetection): 46.1M - Model size (VGG3DDetection) (float): 176 MB | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model |---|---|---|---|---|---|---|---|---| | Yolo2DDetection | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 7.571 ms | 0 - 119 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | Yolo2DDetection | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 7.494 ms | 2 - 118 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | Yolo2DDetection | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 3.836 ms | 0 - 165 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | Yolo2DDetection | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 3.844 ms | 2 - 139 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | Yolo2DDetection | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 1.285 ms | 0 - 2 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | Yolo2DDetection | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 1.283 ms | 0 - 2 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | Yolo2DDetection | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 1.673 ms | 0 - 25 MB | NPU | [3D-Deep-BOX.onnx.zip](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.onnx.zip) | | Yolo2DDetection | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 2.248 ms | 0 - 119 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | Yolo2DDetection | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 2.226 ms | 2 - 119 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | Yolo2DDetection | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 7.571 ms | 0 - 119 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | Yolo2DDetection | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 7.494 ms | 2 - 118 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | Yolo2DDetection | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 3.552 ms | 0 - 127 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | Yolo2DDetection | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 3.492 ms | 0 - 125 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | Yolo2DDetection | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 2.248 ms | 0 - 119 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | Yolo2DDetection | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 2.226 ms | 2 - 119 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | Yolo2DDetection | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 1.033 ms | 0 - 163 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | Yolo2DDetection | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 1.014 ms | 2 - 136 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | Yolo2DDetection | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 1.271 ms | 0 - 109 MB | NPU | [3D-Deep-BOX.onnx.zip](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.onnx.zip) | | Yolo2DDetection | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 0.852 ms | 0 - 125 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | Yolo2DDetection | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 0.845 ms | 2 - 122 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | Yolo2DDetection | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 1.079 ms | 1 - 96 MB | NPU | [3D-Deep-BOX.onnx.zip](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.onnx.zip) | | Yolo2DDetection | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | TFLITE | 0.694 ms | 0 - 123 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | Yolo2DDetection | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | QNN_DLC | 0.706 ms | 2 - 122 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | Yolo2DDetection | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | ONNX | 0.905 ms | 1 - 95 MB | NPU | [3D-Deep-BOX.onnx.zip](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.onnx.zip) | | Yolo2DDetection | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 1.437 ms | 2 - 2 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | Yolo2DDetection | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 1.562 ms | 22 - 22 MB | NPU | [3D-Deep-BOX.onnx.zip](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.onnx.zip) | | VGG3DDetection | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 33.276 ms | 1 - 180 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | VGG3DDetection | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 33.231 ms | 1 - 176 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | VGG3DDetection | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 10.068 ms | 0 - 246 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | VGG3DDetection | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 9.86 ms | 1 - 198 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | VGG3DDetection | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 5.007 ms | 0 - 3 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | VGG3DDetection | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 5.001 ms | 1 - 3 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | VGG3DDetection | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 5.294 ms | 0 - 96 MB | NPU | [3D-Deep-BOX.onnx.zip](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.onnx.zip) | | VGG3DDetection | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 9.178 ms | 0 - 178 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | VGG3DDetection | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 9.176 ms | 1 - 176 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | VGG3DDetection | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 33.276 ms | 1 - 180 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | VGG3DDetection | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 33.231 ms | 1 - 176 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | VGG3DDetection | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 10.07 ms | 0 - 184 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | VGG3DDetection | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 10.077 ms | 1 - 184 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | VGG3DDetection | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 9.178 ms | 0 - 178 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | VGG3DDetection | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 9.176 ms | 1 - 176 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | VGG3DDetection | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 3.868 ms | 0 - 244 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | VGG3DDetection | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 3.882 ms | 1 - 194 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | VGG3DDetection | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 3.972 ms | 0 - 169 MB | NPU | [3D-Deep-BOX.onnx.zip](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.onnx.zip) | | VGG3DDetection | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 3.092 ms | 0 - 186 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | VGG3DDetection | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 3.085 ms | 1 - 182 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | VGG3DDetection | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 3.218 ms | 0 - 152 MB | NPU | [3D-Deep-BOX.onnx.zip](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.onnx.zip) | | VGG3DDetection | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | TFLITE | 2.447 ms | 0 - 182 MB | NPU | [3D-Deep-BOX.tflite](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.tflite) | | VGG3DDetection | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | QNN_DLC | 2.477 ms | 1 - 180 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | VGG3DDetection | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | ONNX | 2.636 ms | 0 - 150 MB | NPU | [3D-Deep-BOX.onnx.zip](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.onnx.zip) | | VGG3DDetection | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 5.097 ms | 1 - 1 MB | NPU | [3D-Deep-BOX.dlc](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.dlc) | | VGG3DDetection | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 5.311 ms | 87 - 87 MB | NPU | [3D-Deep-BOX.onnx.zip](https://huggingface.co/qualcomm/3D-Deep-BOX/blob/main/3D-Deep-BOX.onnx.zip) | ## Installation Install the package via pip: ```bash # NOTE: 3.10 <= PYTHON_VERSION < 3.14 is supported. pip install "qai-hub-models[deepbox]" ``` ## Configure Qualcomm® AI Hub Workbench to run this model on a cloud-hosted device Sign-in to [Qualcomm® AI Hub Workbench](https://workbench.aihub.qualcomm.com/) with your Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`. With this API token, you can configure your client to run models on the cloud hosted devices. ```bash qai-hub configure --api_token API_TOKEN ``` Navigate to [docs](https://workbench.aihub.qualcomm.com/docs/) for more information. ## Demo off target The package contains a simple end-to-end demo that downloads pre-trained weights and runs this model on a sample input. ```bash python -m qai_hub_models.models.deepbox.demo ``` The above demo runs a reference implementation of pre-processing, model inference, and post processing. **NOTE**: If you want running in a Jupyter Notebook or Google Colab like environment, please add the following to your cell (instead of the above). ``` %run -m qai_hub_models.models.deepbox.demo ``` ### Run model on a cloud-hosted device In addition to the demo, you can also run the model on a cloud-hosted Qualcomm® device. This script does the following: * Performance check on-device on a cloud-hosted device * Downloads compiled assets that can be deployed on-device for Android. * Accuracy check between PyTorch and on-device outputs. ```bash python -m qai_hub_models.models.deepbox.export ``` ## How does this work? This [export script](https://aihub.qualcomm.com/models/deepbox/qai_hub_models/models/3D-Deep-BOX/export.py) leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model on-device. Lets go through each step below in detail: Step 1: **Compile model for on-device deployment** To compile a PyTorch model for on-device deployment, we first trace the model in memory using the `jit.trace` and then call the `submit_compile_job` API. ```python import torch import qai_hub as hub from qai_hub_models.models.deepbox import Model # Load the model torch_model = Model.from_pretrained() # Device device = hub.Device("Samsung Galaxy S25") # Trace model input_shape = torch_model.get_input_spec() sample_inputs = torch_model.sample_inputs() pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()]) # Compile model on a specific device compile_job = hub.submit_compile_job( model=pt_model, device=device, input_specs=torch_model.get_input_spec(), ) # Get target model to run on-device target_model = compile_job.get_target_model() ``` Step 2: **Performance profiling on cloud-hosted device** After compiling models from step 1. Models can be profiled model on-device using the `target_model`. Note that this scripts runs the model on a device automatically provisioned in the cloud. Once the job is submitted, you can navigate to a provided job URL to view a variety of on-device performance metrics. ```python profile_job = hub.submit_profile_job( model=target_model, device=device, ) ``` Step 3: **Verify on-device accuracy** To verify the accuracy of the model on-device, you can run on-device inference on sample input data on the same cloud hosted device. ```python input_data = torch_model.sample_inputs() inference_job = hub.submit_inference_job( model=target_model, device=device, inputs=input_data, ) on_device_output = inference_job.download_output_data() ``` With the output of the model, you can compute like PSNR, relative errors or spot check the output with expected output. **Note**: This on-device profiling and inference requires access to Qualcomm® AI Hub Workbench. [Sign up for access](https://myaccount.qualcomm.com/signup). ## Deploying compiled model to Android The models can be deployed using multiple runtimes: - TensorFlow Lite (`.tflite` export): [This tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a guide to deploy the .tflite model in an Android application. - QNN (`.so` export ): This [sample app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html) provides instructions on how to use the `.so` shared library in an Android application. ## View on Qualcomm® AI Hub Get more details on 3D-Deep-BOX's performance across various devices [here](https://aihub.qualcomm.com/models/deepbox). Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/) ## License * The license for the original implementation of 3D-Deep-BOX can be found [here](https://github.com/skhadem/3D-BoundingBox/blob/master/LICENSE). ## References * [3D Bounding Box Estimation Using Deep Learning and Geometry](https://arxiv.org/abs/1612.00496) * [Source Model Implementation](https://github.com/skhadem/3D-BoundingBox/) ## 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).