| cpuClass |
Type of CPU architecture for operation of vehicle and LP detectors. |
- auto - set automatically |
auto |
 |
|
|
- cpu – use CPU |
|
|
|
|
- avx – use AVX2 CPU module |
|
|
| deviceClass |
Type of architecture for the operation of detectors. |
- cpu - Use CPU |
cpu |
 |
|
|
- gpu - Use GPU (cpuClass must be set to auto) |
|
|
| numThreads |
The number of threads on which the detector runs when using the CPU. "-1" value sets the auto-mode for determining the number of threads. |
1...64 |
-1 |
|
| verboseLogging |
The level of logging of the work of neural networks. |
- 0 – logging off; |
0 |
|
|
|
- 1 – logging errors; |
|
|
|
|
- 2 – logging warnings; |
|
|
|
|
- 3 – logging information; |
|
|
|
|
- 4 – logging debug information; |
|
|
|
|
- 5 – logging all. |
|
|
| numComputeStreams |
The number of allocated CUDA threads to run on the GPU architecture. |
1...64 |
4 |
|
| programCacheSize |
Maximum number of program objects in the Flower cache. |
1...9999 |
128 |
 |
| defaultGpuDevice |
The number of the GPU to be used. |
- -1 – do not use GPU; |
0 |
|
|
|
- 0..N – GPU number. |
|
|