I am working on training yolov3-tiny model for custom dataset. Made all the required changes within the .cfg file and caffe.proto files. And I followed the steps as explained by @ChenYingpeng to copy certain caffe layers and remake caffe. But I am facing error while executing "make runtest" as shown below.
[----------] 4 tests from ContrastiveLossLayerTest/1, where TypeParam = caffe::CPUDevice<double>
[ RUN ] ContrastiveLossLayerTest/1.TestGradient
[ OK ] ContrastiveLossLayerTest/1.TestGradient (198 ms)
[ RUN ] ContrastiveLossLayerTest/1.TestGradientLegacy
[ OK ] ContrastiveLossLayerTest/1.TestGradientLegacy (159 ms)
[ RUN ] ContrastiveLossLayerTest/1.TestForward
[ OK ] ContrastiveLossLayerTest/1.TestForward (0 ms)
[ RUN ] ContrastiveLossLayerTest/1.TestForwardLegacy
[ OK ] ContrastiveLossLayerTest/1.TestForwardLegacy (0 ms)
[----------] 4 tests from ContrastiveLossLayerTest/1 (360 ms total)
[----------] 3 tests from MaxPoolingDropoutTest/1, where TypeParam = caffe::CPUDevice<double>
[ RUN ] MaxPoolingDropoutTest/1.TestForward
src/caffe/test/test_maxpool_dropout_layers.cpp:74: Failure
Value of: this->blob_top_->count()
Actual: 72
Expected: sum
Which is: -6.12508e+39
src/caffe/test/test_maxpool_dropout_layers.cpp:85: Failure
Expected: (sum) >= (0), actual: -8.16678e+39 vs 0
[ FAILED ] MaxPoolingDropoutTest/1.TestForward, where TypeParam = caffe::CPUDevice<double> (0 ms)
[ RUN ] MaxPoolingDropoutTest/1.TestBackward
src/caffe/test/test_maxpool_dropout_layers.cpp:110: Failure
Value of: this->blob_top_->count()
Actual: 72
Expected: sum
Which is: 69
double free or corruption (out)
*** Aborted at 1697716999 (unix time) try "date -d @1697716999" if you are using GNU date ***
PC: @ 0x7f604f8bee87 gsignal
*** SIGABRT (@0x3e800001188) received by PID 4488 (TID 0x7f6052c81200) from PID 4488; stack trace: ***
@ 0x7f604fc92980 (unknown)
@ 0x7f604f8bee87 gsignal
@ 0x7f604f8c07f1 abort
@ 0x7f604f909837 (unknown)
@ 0x7f604f9108ba (unknown)
@ 0x7f604f917e4a cfree
@ 0x7f60530ef382 boost::detail::sp_counted_impl_p<>::dispose()
@ 0x7f60530ef78a boost::detail::sp_counted_base::release()
@ 0x7f605327d31a caffe::MaxPoolingDropoutTest<>::~MaxPoolingDropoutTest()
@ 0x7f605327d497 caffe::MaxPoolingDropoutTest_TestBackward_Test<>::~MaxPoolingDropoutTest_TestBackward_Test()
@ 0x7f60533c012a testing::internal::HandleExceptionsInMethodIfSupported<>()
@ 0x7f60533b94ac testing::TestInfo::Run()
@ 0x7f60533b95b5 testing::TestCase::Run()
@ 0x7f60533b9a70 testing::internal::UnitTestImpl::RunAllTests()
@ 0x7f60533b9bb7 testing::UnitTest::Run()
@ 0x7f60530e6e02 main
@ 0x7f604f8a1c87 __libc_start_main
@ 0x7f60530ee32a _start
Aborted (core dumped)
Makefile:542: recipe for target 'runtest' failed
make: *** [runtest] Error 134
If I don't copy the pooling layer and the build caffe I do not get any error. Let me know if any additional modifications are required.
Kindly help me resolve this issue.
Hi @ChenYingpeng ,
I am working on training yolov3-tiny model for custom dataset. Made all the required changes within the .cfg file and caffe.proto files. And I followed the steps as explained by @ChenYingpeng to copy certain caffe layers and remake caffe. But I am facing error while executing "make runtest" as shown below.
If I don't copy the pooling layer and the build caffe I do not get any error. Let me know if any additional modifications are required. Kindly help me resolve this issue.
Thanks in advance.