I’m currently seeking an .onnx model for SDG pallet that supports 3D detection with an image size of 1280x720 pixels. The default model is trained with a smaller image size of 512x512 pixels, and I’m looking for a model that can process larger images. If possible, could you please advise if there is an .onnx model available that has been trained specifically with 1280x720 pixel images? If I use the default model to generate an engine file, it outputs results based on its native 512x512 pixel resolution, which doesn’t meet my current requirements. Any guidance or information you can provide would be greatly appreciated.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| How can I use the dynamic shape of tensorrt ？ | 3 | 2177 | April 8, 2021 | |
| Model file size on jetson nano with 6.0-full-dims is larger than that on desktop PC with 6.0 | 4 | 736 | October 18, 2021 | |
| Dynamic batch size | 3 | 4124 | January 24, 2023 | |
| The same onnx file produces engine files with different size under same configurations | 4 | 671 | May 12, 2022 | |
| Generate Dynamic batch size engine with tensorrt for DLA based CNN Inference | 2 | 40 | September 30, 2024 | |
| MaxPool: at least 5 dimensions are required for input | 3 | 1485 | June 10, 2021 | |
| Failing custom object detection | 6 | 512 | January 5, 2022 | |
| Create Ses image out of engine output | 1 | 469 | April 21, 2022 | |
| Where I can find onnx model of RetinaNet? | 0 | 659 | September 9, 2019 | |
| assertion failed: onnx_padding.size() == 8 | 3 | 1044 | October 12, 2021 |