mirror/opencv
1
0
Fork 0
mirror of https://github.com/opencv/opencv.git synced 2024-11-27 20:50:25 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

restore cafe-specific tests

Browse Source
This commit is contained in:
Wanli 2024-08-21 03:57:54 +08:00
parent d231b4e362
commit 7e4a326ba4
2 changed files with 40 additions and 25 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
modules/dnn/misc/python/test/test_dnn.py
View File

@ -110,9 +110,8 @@ class dnn_test(NewOpenCVTests):
required=required)
def checkIETarget(self, backend, target):
proto = self.find_dnn_file('dnn/layers/layer_convolution.prototxt')
model = self.find_dnn_file('dnn/layers/layer_convolution.caffemodel')
net = cv.dnn.readNet(proto, model)
model = self.find_dnn_file('dnn/layers/layer_convolution.onnx')
net = cv.dnn.readNet(model)
try:
net.setPreferableBackend(backend)
net.setPreferableTarget(target)
@ -324,10 +323,9 @@ class dnn_test(NewOpenCVTests):
def test_async(self):
timeout = 10*1000*10**6 # in nanoseconds (10 sec)
proto = self.find_dnn_file('dnn/layers/layer_convolution.prototxt')
model = self.find_dnn_file('dnn/layers/layer_convolution.caffemodel')
if proto is None or model is None:
raise unittest.SkipTest("Missing DNN test files (dnn/layers/layer_convolution.{prototxt/caffemodel}). Verify OPENCV_DNN_TEST_DATA_PATH configuration parameter.")
model = self.find_dnn_file('dnn/layers/layer_convolution.onnx')
if model is None:
raise unittest.SkipTest("Missing DNN test files (dnn/layers/layer_convolution.onnx). Verify OPENCV_DNN_TEST_DATA_PATH configuration parameter.")
print('\n')
for backend, target in self.dnnBackendsAndTargets:
@ -336,11 +334,11 @@ class dnn_test(NewOpenCVTests):
printParams(backend, target)
netSync = cv.dnn.readNet(proto, model)
netSync = cv.dnn.readNet(model)
netSync.setPreferableBackend(backend)
netSync.setPreferableTarget(target)
netAsync = cv.dnn.readNet(proto, model)
netAsync = cv.dnn.readNet(model)
netAsync.setPreferableBackend(backend)
netAsync.setPreferableTarget(target)

49
modules/dnn/test/test_layers.cpp
View File

@ -65,13 +65,26 @@ static String _tf(TString filename)
class Test_Caffe_layers : public DNNTestLayer
{
public:
void testLayerUsingOnnxModels(const String& basename, bool useCommonInputBlob = true, double l1 = 0.0, double lInf = 0.0,
int numInps = 1, int numOuts = 1)
{
_testLayerUsingModels(basename, false, true, useCommonInputBlob, l1, lInf, numInps, numOuts);
}
void testLayerUsingCaffeModels(const String& basename, bool useCaffeModel = false,
bool useCommonInputBlob = true, double l1 = 0.0, double lInf = 0.0,
int numInps = 1, int numOuts = 1)
{
_testLayerUsingModels(basename, useCaffeModel, false, useCommonInputBlob, l1, lInf, numInps, numOuts);
}
void _testLayerUsingModels(const String& basename, bool useCaffeModel, bool useOnnxModel, bool useCommonInputBlob, double l1, double lInf,
int numInps, int numOuts)
{
CV_Assert_N(numInps >= 1, numInps <= 10, numOuts >= 1, numOuts <= 10);
String prototxt = _tf(basename + ".prototxt");
String caffemodel = _tf(basename + ".caffemodel");
String onnxmodel = _tf(basename + ".onnx");
std::vector<Mat> inps, refs, outs;
@ -103,7 +116,11 @@ public:
refs.push_back(blobFromNPY(outfile));
}
Net net = readNetFromCaffe(prototxt, (useCaffeModel) ? caffemodel : String());
Net net;
if (useOnnxModel)
net = readNetFromONNX(onnxmodel);
else
net = readNetFromCaffe(prototxt, (useCaffeModel) ? caffemodel : String());
ASSERT_FALSE(net.empty());
checkBackend(&inps[0], &refs[0]);
@ -133,7 +150,7 @@ public:
TEST_P(Test_Caffe_layers, Softmax)
{
testLayerUsingCaffeModels("layer_softmax");
testLayerUsingOnnxModels("layer_softmax");
}
TEST_P(Test_Caffe_layers, LRN)
@ -148,12 +165,12 @@ TEST_P(Test_Caffe_layers, LRN)
lInf = 0.01;
}
testLayerUsingCaffeModels("layer_lrn_spatial", false, true, l1, lInf);
testLayerUsingCaffeModels("layer_lrn_channels", false, true, l1, lInf);
testLayerUsingOnnxModels("layer_lrn_channels", true, l1, lInf);
}
TEST_P(Test_Caffe_layers, Convolution)
{
testLayerUsingCaffeModels("layer_convolution", true);
testLayerUsingOnnxModels("layer_convolution", true);
}
TEST_P(Test_Caffe_layers, DeConvolution)
@ -195,12 +212,12 @@ TEST_P(Test_Caffe_layers, InnerProduct)
TEST_P(Test_Caffe_layers, Pooling_max)
{
testLayerUsingCaffeModels("layer_pooling_max");
testLayerUsingOnnxModels("layer_pooling_max");
}
TEST_P(Test_Caffe_layers, Pooling_ave)
{
testLayerUsingCaffeModels("layer_pooling_ave");
testLayerUsingOnnxModels("layer_pooling_ave");
}
TEST_P(Test_Caffe_layers, MVN)
@ -260,18 +277,18 @@ TEST(Layer_Test_Reshape, Accuracy)
TEST_P(Test_Caffe_layers, BatchNorm)
{
testLayerUsingCaffeModels("layer_batch_norm", true);
testLayerUsingOnnxModels("layer_batch_norm", true);
testLayerUsingCaffeModels("layer_batch_norm_local_stats", true, false);
}
TEST_P(Test_Caffe_layers, ReLU)
{
testLayerUsingCaffeModels("layer_relu");
testLayerUsingOnnxModels("layer_relu");
}
TEST_P(Test_Caffe_layers, Dropout)
{
testLayerUsingCaffeModels("layer_dropout");
testLayerUsingOnnxModels("layer_dropout");
}
TEST_P(Test_Caffe_layers, Concat)
@ -295,9 +312,9 @@ TEST_P(Test_Caffe_layers, Concat)
#endif
#endif
testLayerUsingCaffeModels("layer_concat");
testLayerUsingCaffeModels("layer_concat_optim", true, false);
testLayerUsingCaffeModels("layer_concat_shared_input", true, false);
testLayerUsingOnnxModels("layer_concat");
testLayerUsingOnnxModels("layer_concat_optim", false);
testLayerUsingOnnxModels("layer_concat_shared_input", false);
}
TEST_P(Test_Caffe_layers, Fused_Concat)
@ -356,7 +373,7 @@ TEST_P(Test_Caffe_layers, Eltwise)
TEST_P(Test_Caffe_layers, PReLU)
{
double lInf = (target == DNN_TARGET_MYRIAD || target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_CPU_FP16) ? 0.021 : 0.0;
testLayerUsingCaffeModels("layer_prelu", true, true, 0.0, lInf);
testLayerUsingOnnxModels("layer_prelu", true, 0.0, lInf);
}
// TODO: fix an unstable test case
@ -377,7 +394,7 @@ TEST_P(Test_Caffe_layers, layer_prelu_fc)
l1 = 0.01f; lInf = 0.05f;
}
#endif
testLayerUsingCaffeModels("layer_prelu_fc", true, false, l1, lInf);
testLayerUsingOnnxModels("layer_prelu_fc", false, l1, lInf);
}
TEST_P(Test_Caffe_layers, Reshape_Split_Slice)
@ -756,7 +773,7 @@ TEST_P(Test_Caffe_layers, ChannelNorm)
{
if (backend == DNN_BACKEND_OPENCV && target == DNN_TARGET_OPENCL_FP16)
applyTestTag(CV_TEST_TAG_DNN_SKIP_OPENCL_FP16);
testLayerUsingCaffeModels("channel_norm", false, false);
testLayerUsingOnnxModels("channel_norm", false);
}
TEST_P(Test_Caffe_layers, DataAugmentation)
@ -1265,7 +1282,7 @@ TEST_P(Layer_Test_Convolution_DLDT, Accuracy)
ASSERT_EQ(DNN_BACKEND_INFERENCE_ENGINE_NGRAPH, backendId);
Net netDefault = readNet(_tf("layer_convolution.caffemodel"), _tf("layer_convolution.prototxt"));
Net netDefault = readNet(_tf("layer_convolution.onnx"));
Net net = readNet(_tf("layer_convolution.xml"), _tf("layer_convolution.bin"));
Mat inp = blobFromNPY(_tf("blob.npy"));