mx/Jungfraujoch
0
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases 37 Wiki Activity

Add deep learning resolution estimation model from Stanford

Browse Source
This commit is contained in:
leonarski_f
2024-02-08 20:15:29 +01:00
parent 1b4ab88f54
commit 8dcecb9685
49 changed files with 650 additions and 194 deletions
Download Patch File Download Diff File Expand all files Collapse all files

53
tests/NeuralNetResPredictorTest.cpp Normal file
View File

@@ -0,0 +1,53 @@
// Copyright (2019-2024) Paul Scherrer Institute
#include <catch2/catch.hpp>
#include "../writer/HDF5Objects.h"
#include "../resonet/NeuralNetResPredictor.h"
TEST_CASE("NeuralNetResPredictor_Prepare", "[LinearAlgebra][Coord]") {
DiffractionExperiment experiment(DetectorGeometry(8, 2, 8, 36));
experiment.DetectorDistance_mm(75).PhotonEnergy_keV(12.4).BeamX_pxl(1000).BeamY_pxl(1000);
experiment.NeuralNetModelPath("../../resonet/traced_resnet_model.pt");
std::vector<int16_t> v(experiment.GetPixelsNum(),0);
v[1000 * experiment.GetXPixelsNum() + 1000] = 100;
v[1000 * experiment.GetXPixelsNum() + 1001] = 20;
v[1001 * experiment.GetXPixelsNum() + 1000] = 30;
v[1001 * experiment.GetXPixelsNum() + 1001] = INT16_MIN;
v[1050 * experiment.GetXPixelsNum() + 1050] = 52;
v[2000 * experiment.GetXPixelsNum() + 1500] = 160;
NeuralNetResPredictor predictor(experiment);
REQUIRE(predictor.GetMaxPoolFactor() == 2);
predictor.Prepare(v.data());
auto nn_input = predictor.GetModelInput();
REQUIRE(nn_input[0] == 10);
REQUIRE(nn_input[25 * 512 + 25] == 7);
REQUIRE(nn_input[500 * 512 + 250] == 12);
}
TEST_CASE("NeuralNetResPredictor_Inference", "[LinearAlgebra][Coord]") {
DiffractionExperiment experiment(DetectorGeometry(8, 2, 8, 36));
experiment.DetectorDistance_mm(75).PhotonEnergy_keV(12.4).BeamY_pxl(1136).BeamX_pxl(1090);
experiment.NeuralNetModelPath("../../resonet/traced_resnet_model.pt");
NeuralNetResPredictor predictor(experiment);
HDF5ReadOnlyFile data("../../tests/test_data/compression_benchmark.h5");
HDF5DataSet dataset(data, "/entry/data/data");
HDF5DataSpace file_space(dataset);
std::vector<int16_t> image_conv (file_space.GetDimensions()[1] * file_space.GetDimensions()[2]);
std::vector<hsize_t> start = {4,0,0};
std::vector<hsize_t> file_size = {1, file_space.GetDimensions()[1], file_space.GetDimensions()[2]};
dataset.ReadVector(image_conv, start, file_size);
auto res = predictor.Inference(image_conv.data());
std::cout << res << std::endl;
REQUIRE(res < 1.5);
REQUIRE(res > 1.4);
}