optimize face recognizer for dual_buff mode

This commit is contained in:
Neucrack
2024-07-16 18:55:27 +08:00
parent 1174431e69
commit f3b98299a7

View File

@@ -319,7 +319,7 @@ namespace maix::nn
image::Image *std_img = img.affine(objs->at(i).points, _std_points, _feature_input_size, _feature_input_size);
// img.save("/root/test0.jpg");
// std_img->save("/root/test.jpg");
tensor::Tensors *outputs = _model_feature->forward_image(*std_img, this->mean_feature, this->scale_feature, fit, false);
tensor::Tensors *outputs = _model_feature->forward_image(*std_img, this->mean_feature, this->scale_feature, fit, false, get_feature || get_face);
if (!outputs) // not ready for dual_buff mode
{
delete std_img;