you can run the draw_bbox.py to visualize the 3d data of the dataset. Open3D is required for the visualization.
There are totally 9 class in the dataset.
distance=(1-depth_value/255.0)*10
For both 3D object detection and depth estimation, submit your results (zip it if needed), output_report.json generated by the evaluation
script, and your codes and model. Make sure we can run your code from the instruction that you provide. Send all above
files to our email: [email protected]. Maxmimum submission is once per day.
For the result on the test dataset. Please use only the rgb image as inputs to generate results. Please also submit the raw result on test dataset to our email, which should be able to pass the evaluation script.
Team | CIE | Friend | 天气太热了 | 以此为归 | Inceptio | 万里蹀躞 | NH106A | 全都 |
---|---|---|---|---|---|---|---|---|
rmse | 0.280860 | 0.305452 | 0.307599 | 0.307626 | 0.325035 | 0.336830 | 1.428464 | 2.106284 |
Team | CIE | 以此为归 | 天气太热了 | Friend | Inceptio | 万里蹀躞 |
---|---|---|---|---|---|---|
rmse | 0.332012 | 0.351462 | 0.352469 | 0.355269 | 0.366891 | 0.369997 |