Support of inference with TensorRT for sshaoshuai/MTR.
trajectory <float; [B, N, Tp, Da]>
trajectory_mask <bool: [B, N, Tp]>
polyline <float: [B, K, P, Dp]>
polyline_mask <bool: [B, K, P]>
polyline_center <float: [B, K, 3]>
last_pos <float; [B, N, 3]
track_index <int: [B]>
label_index <int: [N]>
intention_points <float: [B, 64, 3]>
where,
B
...The number of target agentsN
...The number of all agentsTp
...The number of past frames(=11
)Da
...The number of agent state dimensions(=29
)K
...The max number of polylines(=768
)P
...The max number of points contained in each polyline(=20
)Dp
...The number of polyline state dimensions(=9
)
scores <float: [B, M]>
trajectory <float: [B, M, Tf, Dt]>
where,
M
...The number of modesTf
...The number of the predicted future frames(=80
)Dt
...The number of the predicted trajectory dimensions(=7
) in the order of(x, y, dx, dy, yaw, vx, cy)
.
cmake -B build
cmake --build build -j${nproc}
# download onnx.zip
git lfs pull
# with trtexec
<PATH_TO_TRTEXEC_BIN>/trtexec --onnx=<PATH_TO_ONNX> --staticPlugins=./build/libcustom_plugin.so
# main
./build/main <PATH_TO_ONNX_OR_ENGINE> [--dynamic --fp16 -n <NUM_REPEAT>]
# test agent data container defined in `include/mtr/agent.hpp`
./build/test_agent
# test polyline data container defined in `include/mtr/polyline.hpp`
./build/test_polyline
# test intention point data container defined in `include/mtr/intention_point.hpp`
./build/test_intention_point
- TensorRT custom plugins
- CUDA kernels
- pre-process
- post-process
- Shape inference
- static shape
- dynamic shape
- Inference sample
- Visualization
- Evaluation