--- library_name: pytorch license: agpl-3.0 pipeline_tag: object-detection tags: - real_time - quantized - android --- ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/yolov8_det_quantized/web-assets/model_demo.png) # YOLOv8-Detection-Quantized: Optimized for Mobile Deployment ## Quantized real-time object detection optimized for mobile and edge by Ultralytics Ultralytics YOLOv8 is a machine learning model that predicts bounding boxes and classes of objects in an image. This model is post-training quantized to int8 using samples from the COCO dataset. This model is an implementation of YOLOv8-Detection-Quantized found [here](https://github.com/ultralytics/ultralytics/tree/main/ultralytics/models/yolo/detect). More details on model performance across various devices, can be found [here](https://aihub.qualcomm.com/models/yolov8_det_quantized). ### Model Details - **Model Type:** Object detection - **Model Stats:** - Model checkpoint: YOLOv8-N - Input resolution: 640x640 - Number of parameters: 3.18M - Model size: 3.26 MB | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model |---|---|---|---|---|---|---|---|---| | YOLOv8-Detection-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 1.909 ms | 0 - 10 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 2.279 ms | 1 - 12 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 6.413 ms | 0 - 31 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 1.286 ms | 0 - 28 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 1.532 ms | 1 - 41 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 4.534 ms | 1 - 55 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 1.214 ms | 0 - 28 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 1.476 ms | 1 - 32 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 4.064 ms | 0 - 49 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | RB3 Gen 2 (Proxy) | QCS6490 Proxy | TFLITE | 4.743 ms | 0 - 30 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | RB3 Gen 2 (Proxy) | QCS6490 Proxy | QNN | 5.967 ms | 1 - 12 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | RB5 (Proxy) | QCS8250 Proxy | TFLITE | 49.035 ms | 3 - 12 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 1.921 ms | 0 - 13 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 2.038 ms | 1 - 4 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | SA7255P ADP | SA7255P | TFLITE | 11.668 ms | 0 - 22 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | SA7255P ADP | SA7255P | QNN | 12.004 ms | 1 - 10 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 1.919 ms | 0 - 6 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | SA8255 (Proxy) | SA8255P Proxy | QNN | 2.036 ms | 1 - 3 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | SA8295P ADP | SA8295P | TFLITE | 2.833 ms | 0 - 29 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | SA8295P ADP | SA8295P | QNN | 3.225 ms | 1 - 16 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 1.911 ms | 0 - 10 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | SA8650 (Proxy) | SA8650P Proxy | QNN | 2.036 ms | 1 - 4 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | SA8775P ADP | SA8775P | TFLITE | 2.871 ms | 0 - 22 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | SA8775P ADP | SA8775P | QNN | 3.109 ms | 1 - 11 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 2.123 ms | 0 - 34 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 2.556 ms | 1 - 34 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 2.29 ms | 1 - 1 MB | INT8 | NPU | -- | | YOLOv8-Detection-Quantized | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 6.772 ms | 5 - 5 MB | INT8 | NPU | -- | ## License * The license for the original implementation of YOLOv8-Detection-Quantized can be found [here](https://github.com/ultralytics/ultralytics/blob/main/LICENSE). * The license for the compiled assets for on-device deployment can be found [here](https://github.com/ultralytics/ultralytics/blob/main/LICENSE) ## References * [Ultralytics YOLOv8 Docs: Object Detection](https://docs.ultralytics.com/tasks/detect/) * [Source Model Implementation](https://github.com/ultralytics/ultralytics/tree/main/ultralytics/models/yolo/detect) ## Community * Join [our AI Hub Slack community](https://qualcomm-ai-hub.slack.com/join/shared_invite/zt-2d5zsmas3-Sj0Q9TzslueCjS31eXG2UA#/shared-invite/email) 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). ## Usage and Limitations Model may not be used for or in connection with any of the following applications: - Accessing essential private and public services and benefits; - Administration of justice and democratic processes; - Assessing or recognizing the emotional state of a person; - Biometric and biometrics-based systems, including categorization of persons based on sensitive characteristics; - Education and vocational training; - Employment and workers management; - Exploitation of the vulnerabilities of persons resulting in harmful behavior; - General purpose social scoring; - Law enforcement; - Management and operation of critical infrastructure; - Migration, asylum and border control management; - Predictive policing; - Real-time remote biometric identification in public spaces; - Recommender systems of social media platforms; - Scraping of facial images (from the internet or otherwise); and/or - Subliminal manipulation