Crops rectangular regions from the input frames and rescales them all to the same dimensions. Optionally, can also perform color conversion from NV12 to interleaved or planar RGB(A) formats. The inputs are an array of input frames and an array of DCF tracked bounding boxes, specifying the corresponding frame index for each object, the starting coordinates, width and height in pixels, and tracking state (lost object are ignored). The output is a single image where all output patches are stacked vertically.
The input is sampled using bilinear interpolation. If the bounding box is partially out of frame, clamping is used (the border pixels are repeated).
For list of limitations, constraints and backends that implement the algorithm, consult reference documentation of the following functions:
| Function | Description |
|---|---|
| vpiCreateCropScaler | Creates payload instance for the Crop Scale algorithm. |
| vpiSubmitCropScalerBatch | Crops rectangular regions from the input frames and rescale them all to the same dimensions. |
For more information, see Crop Scaler in the "C API Reference" section of VPI - Vision Programming Interface.
For information on how to use the performance tables below, see Algorithm Performance Tables.
Before comparing measurements, consult Comparing Algorithm Elapsed Times.
For further information on how performance was benchmarked, see Performance Benchmark.