The project refers to AdasSourrondView[3], modified to a cuda-accelerated version.
index | repo | info |
---|---|---|
1 | surround-view-system-introduction | python verison, refrence repo for 2D avm |
2 | 3d surround-view-system | cuda+opengl verison, for 3d avm |
3 | AdasSourrondView | c++ verison |
run surround_view.ipynb or nvcc cuda.cu -o cuda -lopencv_core -lopencv_imgcodecs && ./cuda
The code inputs are 4 fisheye camera images, and the stitched image results are as follows. The project utilizes cuda to accelerate the c++ code with a frame rate over 10k.