metadata
library_name: pytorch
license: gpl-3.0
tags:
- real_time
- android
pipeline_tag: object-detection
Yolo-v7: Optimized for Mobile Deployment
Real-time object detection optimized for mobile and edge
YoloV7 is a machine learning model that predicts bounding boxes and classes of objects in an image.
This model is an implementation of Yolo-v7 found here.
More details on model performance across various devices, can be found here.
Model Details
- Model Type: Object detection
- Model Stats:
- Model checkpoint: YoloV7 Tiny
- Input resolution: 640x640
- Number of parameters: 6.39M
- Model size: 24.4 MB
| Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model |
|---|---|---|---|---|---|---|---|---|
| Yolo-v7 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 15.365 ms | 1 - 18 MB | FP16 | NPU | -- |
| Yolo-v7 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 10.581 ms | 5 - 7 MB | FP16 | NPU | -- |
| Yolo-v7 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 12.474 ms | 2 - 62 MB | FP16 | NPU | -- |
| Yolo-v7 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 10.257 ms | 0 - 46 MB | FP16 | NPU | -- |
| Yolo-v7 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 6.927 ms | 5 - 24 MB | FP16 | NPU | -- |
| Yolo-v7 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 8.417 ms | 7 - 73 MB | FP16 | NPU | -- |
| Yolo-v7 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 10.701 ms | 1 - 45 MB | FP16 | NPU | -- |
| Yolo-v7 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 7.098 ms | 5 - 72 MB | FP16 | NPU | -- |
| Yolo-v7 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 8.23 ms | 5 - 63 MB | FP16 | NPU | -- |
| Yolo-v7 | SA7255P ADP | SA7255P | TFLITE | 107.906 ms | 1 - 38 MB | FP16 | NPU | -- |
| Yolo-v7 | SA7255P ADP | SA7255P | QNN | 100.592 ms | 0 - 7 MB | FP16 | NPU | -- |
| Yolo-v7 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 15.294 ms | 1 - 22 MB | FP16 | NPU | -- |
| Yolo-v7 | SA8255 (Proxy) | SA8255P Proxy | QNN | 10.44 ms | 5 - 7 MB | FP16 | NPU | -- |
| Yolo-v7 | SA8295P ADP | SA8295P | TFLITE | 19.717 ms | 1 - 41 MB | FP16 | NPU | -- |
| Yolo-v7 | SA8295P ADP | SA8295P | QNN | 13.339 ms | 0 - 11 MB | FP16 | NPU | -- |
| Yolo-v7 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 15.172 ms | 1 - 19 MB | FP16 | NPU | -- |
| Yolo-v7 | SA8650 (Proxy) | SA8650P Proxy | QNN | 10.411 ms | 5 - 7 MB | FP16 | NPU | -- |
| Yolo-v7 | SA8775P ADP | SA8775P | TFLITE | 20.463 ms | 1 - 39 MB | FP16 | NPU | -- |
| Yolo-v7 | SA8775P ADP | SA8775P | QNN | 14.809 ms | 1 - 8 MB | FP16 | NPU | -- |
| Yolo-v7 | QCS8275 (Proxy) | QCS8275 Proxy | TFLITE | 107.906 ms | 1 - 38 MB | FP16 | NPU | -- |
| Yolo-v7 | QCS8275 (Proxy) | QCS8275 Proxy | QNN | 100.592 ms | 0 - 7 MB | FP16 | NPU | -- |
| Yolo-v7 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 15.317 ms | 1 - 22 MB | FP16 | NPU | -- |
| Yolo-v7 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 10.404 ms | 5 - 7 MB | FP16 | NPU | -- |
| Yolo-v7 | QCS9075 (Proxy) | QCS9075 Proxy | TFLITE | 20.463 ms | 1 - 39 MB | FP16 | NPU | -- |
| Yolo-v7 | QCS9075 (Proxy) | QCS9075 Proxy | QNN | 14.809 ms | 1 - 8 MB | FP16 | NPU | -- |
| Yolo-v7 | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 17.742 ms | 1 - 52 MB | FP16 | NPU | -- |
| Yolo-v7 | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 12.669 ms | 5 - 63 MB | FP16 | NPU | -- |
| Yolo-v7 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 10.965 ms | 5 - 5 MB | FP16 | NPU | -- |
| Yolo-v7 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 14.241 ms | 10 - 10 MB | FP16 | NPU | -- |
License
- The license for the original implementation of Yolo-v7 can be found here.
- The license for the compiled assets for on-device deployment can be found here
References
- YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors
- Source Model Implementation
Community
- Join our AI Hub Slack community to collaborate, post questions and learn more about on-device AI.
- For questions or feedback please reach out to us.
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