Skip to content
Log in

fundamentalvision/BEVFormer

Embedl's build embedl/bevformer-tiny of the original model fundamentalvision/BEVFormer

fp16
BEVFormer-tiny architecture graph
BEVFormer-tiny architecture graph
  • fundamentalvision/BEVFormer
  • FP16
  • AGX Orin
Labels
  • Image
  • Detection
Build
Baseline
Precision
FP16
Runtime
TensorRT 8.5.1.7 with BEVFormer_tensorrt plugins
Cameras
6 views, 800×480 each
BEV grid
50×50

BEVFormer-tiny (R50 backbone) detects 3D objects in the ten nuScenes classes from six surround-view cameras.

BEVFormer is by Zhiqi Li et al. (arXiv:2203.17270), Apache-2.0, at fundamentalvision/BEVFormer; the TensorRT export and plugins are from DerryHub/BEVFormer_tensorrt, Apache-2.0.

Files

Artifacts

What ships in the model bundle.

File Kind Size sha256
TensorRT engine

engine/bevformer_tiny_trt_p_cp_trtexec.engine

 Engine 69.7 MiB 07d616a7dc11
trtexec build log

logs/trtexec_build.log

 Log 9.9 MiB 1938d161d792
HTTP server for the engine (serve.sh --port 8000)

engine/serve.sh

 script 2 KiB c59eaebcab93
Model settings

bevformer.json

 Config 396 B 2f2efe384667
Latency (trtexec)

benchmark/metrics.json

 Results 4 KiB 519448f1b08b
Accuracy (PyTorch, on the host)

benchmark/accuracy.json

 Results 269 B cda0e1da1161
nuScenes sample (engine inputs, .npz)

inference/nuscenes_sample.npz

 sample 28.8 MiB 028039f51765
Detections for the sample

inference/detections.json

 Results 101 KiB 4a35ec92c614
BEVFormer_tensorrt licence

LICENSE-BEVFormer_tensorrt

 license 11 KiB c71d239df917