GpuVeryFastIntegralChannelsDetector

Issue #80 new
Anonymous created an issue

My System

OS: Ubuntu 14.10

OpenCV: 2.4.9

NVIDIA Driver: 346.47

nvcc: V6.5.12

My Problem

I successfully compiled ground_estimation after lots of fiddling to get the correct versions of OpenCV, Cuda, Protobuf, Gcc, and the Nvidia Driver (note that the current versions provided via package managers were not friendly with doppia on my machine). I had to reinstall all of them manually.

I'm now moving on to objects_detection (and eventually trying to use HeadHunter). I get all the way up to this command befor crashing:

Command

OMP_NUM_THREADS=4 ./objects_detection -c cvpr2012_very_fast_over_bahnhof.config.ini --gui.disable false

Output w/ Error

Stereo objects detection on a flat world. Rodrigo Benenson, Markus Mathias, Mohamed Omran @ KULeuven & MPI-Inf. 2011-2014.
Parsed the configuration file cvpr2012_very_fast_over_bahnhof.config.ini
Using stereo camera calibration file: ../../video_input/calibration/stereo_calibration_bahnhof.proto.txt
2015-03-14 20:40:31 {7f0ddcf85ac0} [ StereoCameraCalibration ] : stereo_calibration_data name: Andreas Ess Bahnhofstrasse sequence stereo calibration
Reading files:
../../../data/sample_test_images/bahnhof/image_00000000_0.png
../../../data/sample_test_images/bahnhof/image_00000000_1.png
libprotobuf ERROR google/protobuf/message_lite.cc:123] Can't parse message of type "doppia_protobuf.DetectorModel" because it is missing required fields: detector_type
2015-03-14 20:40:31 {7f0ddcf85ac0} [ ObjectsDetectorFactory ] : Model file ../../../data/trained_models/2012_04_04_1417_trained_model_multiscales_synthetic_softcascade.proto.bin is not in binary protocol buffer format
libprotobuf ERROR google/protobuf/text_format.cc:172] Error parsing text-format doppia_protobuf.DetectorModel: 2:1: Invalid control characters encountered in text.
libprotobuf ERROR google/protobuf/text_format.cc:172] Error parsing text-format doppia_protobuf.DetectorModel: 2:9: Message type "doppia_protobuf.DetectorModel" has no field named "Scales".
2015-03-14 20:40:31 {7f0ddcf85ac0} [ ObjectsDetectorFactory ] : Model file ../../../data/trained_models/2012_04_04_1417_trained_model_multiscales_synthetic_softcascade.proto.bin is not in text protocol buffer format
2015-03-14 20:40:31 {7f0ddcf85ac0} [ ObjectsDetectorFactory ] : Parsed the binary model file ../../../data/trained_models/2012_04_04_1417_trained_model_multiscales_synthetic_softcascade.proto.bin
2015-03-14 20:40:31 {7f0ddcf85ac0} [ ObjectsDetectorFactory ] : Parsed the binary model file ../../../data/trained_models/2012_04_04_1417_trained_model_multiscales_synthetic_softcascade.proto.bin
add_cascade_offset: at stage 0 the offset value is -0.05000
add_cascade_offset: at stage 1000 the offset value is -0.05000
add_cascade_offset: at stage 1999 the offset value is -0.05000
add_cascade_offset: at stage 0 the offset value is -0.05000
add_cascade_offset: at stage 1000 the offset value is -0.05000
add_cascade_offset: at stage 1999 the offset value is -0.05000
add_cascade_offset: at stage 0 the offset value is -0.05000
add_cascade_offset: at stage 1000 the offset value is -0.05000
add_cascade_offset: at stage 1999 the offset value is -0.05000
add_cascade_offset: at stage 0 the offset value is -0.05000
add_cascade_offset: at stage 1000 the offset value is -0.05000
add_cascade_offset: at stage 1999 the offset value is -0.05000
add_cascade_offset: at stage 0 the offset value is -0.05000
add_cascade_offset: at stage 1000 the offset value is -0.05000
add_cascade_offset: at stage 1999 the offset value is -0.05000
2015-03-14 20:40:31 {7f0ddcf85ac0} [ MultiScalesIntegralChannelsModel ] : Parsing multiscales model Scales 0.5, 1.0, 2.0, 4.0, 8.0
2015-03-14 20:40:31 {7f0ddcf85ac0} [ MultiScalesIntegralChannelsModel ] : Multiscales/bundle model is being normalized to have maximum detection score == 1
2015-03-14 20:40:31 {7f0ddcf85ac0} [ MultiScalesIntegralChannelsModel ] : Model index 0 has scale 0.5
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : Detector model is being normalized to have maximum detection score == 1
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : The soft cascade contains 2000 stages
Stage 0 cascade_threshold == -0.010148
Stage 999 cascade_threshold == -0.026132
Stage 1999 cascade_threshold == -0.010174
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : Warning: No semantic category found in model, assuming a pedestrian detector
2015-03-14 20:40:31 {7f0ddcf85ac0} [ check_stages_and_range_visitor ] : Warning: Scale -1 is almost fine, min (x, y) == (0, 0), max (x,y) == (7, 15), width, height == (8, 16), (only touches left, top, border(s))
2015-03-14 20:40:31 {7f0ddcf85ac0} [ check_stages_and_range_visitor ] : Warning: Not touching the borders indicates that the model is suboptimal regarding detection quality (was trained with doppia v1, instead of v2)
Single scale/occlusion model sanity check passed.
2015-03-14 20:40:31 {7f0ddcf85ac0} [ MultiScalesIntegralChannelsModel ] : Model index 1 has scale 1
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : Detector model is being normalized to have maximum detection score == 1
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : The soft cascade contains 2000 stages
Stage 0 cascade_threshold == -0.010119
Stage 999 cascade_threshold == -0.026103
Stage 1999 cascade_threshold == -0.010146
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : Warning: No semantic category found in model, assuming a pedestrian detector
2015-03-14 20:40:31 {7f0ddcf85ac0} [ check_stages_and_range_visitor ] : Warning: Scale -1 is almost fine, min (x, y) == (0, 0), max (x,y) == (15, 31), width, height == (16, 32), (only touches left, top, border(s))
Single scale/occlusion model sanity check passed.
2015-03-14 20:40:31 {7f0ddcf85ac0} [ MultiScalesIntegralChannelsModel ] : Model index 2 has scale 2
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : Detector model is being normalized to have maximum detection score == 1
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : The soft cascade contains 2000 stages
Stage 0 cascade_threshold == -0.010106
Stage 999 cascade_threshold == -0.026090
Stage 1999 cascade_threshold == -0.010132
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : Warning: No semantic category found in model, assuming a pedestrian detector
2015-03-14 20:40:31 {7f0ddcf85ac0} [ check_stages_and_range_visitor ] : Warning: Scale -1 is almost fine, min (x, y) == (0, 0), max (x,y) == (31, 63), width, height == (32, 64), (only touches left, top, border(s))
Single scale/occlusion model sanity check passed.
2015-03-14 20:40:31 {7f0ddcf85ac0} [ MultiScalesIntegralChannelsModel ] : Model index 3 has scale 4
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : Detector model is being normalized to have maximum detection score == 1
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : The soft cascade contains 2000 stages
Stage 0 cascade_threshold == -0.010108
Stage 999 cascade_threshold == -0.026092
Stage 1999 cascade_threshold == -0.010135
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : Warning: No semantic category found in model, assuming a pedestrian detector
2015-03-14 20:40:31 {7f0ddcf85ac0} [ check_stages_and_range_visitor ] : Warning: Scale -1 is almost fine, min (x, y) == (0, 0), max (x,y) == (63, 127), width, height == (64, 128), (only touches left, top, border(s))
Single scale/occlusion model sanity check passed.
2015-03-14 20:40:31 {7f0ddcf85ac0} [ MultiScalesIntegralChannelsModel ] : Model index 4 has scale 8
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : Detector model is being normalized to have maximum detection score == 1
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : The soft cascade contains 2000 stages
Stage 0 cascade_threshold == -0.010115
Stage 999 cascade_threshold == -0.026099
Stage 1999 cascade_threshold == -0.010142
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : Warning: No semantic category found in model, assuming a pedestrian detector
2015-03-14 20:40:31 {7f0ddcf85ac0} [ check_stages_and_range_visitor ] : Warning: Scale -1 is almost fine, min (x, y) == (0, 0), max (x,y) == (127, 255), width, height == (128, 256), (only touches left, top, border(s))
Single scale/occlusion model sanity check passed.
2015-03-14 20:40:31 {7f0ddcf85ac0} [ SoftCascadeOverIntegralChannelsModel ] : The detector model seems to include a cascade thresholds
2015-03-14 20:40:31 {7f0ddcf85ac0} [ ObjectsDetectorFactory ] : Updating the score threshold using the last cascade threshold of detector 0. New score threshold == -0.010 (= 0.000 + -0.010)
A std::exception was raised: This executable was compiled without support for GpuVeryFastIntegralChannelsDetector
terminate called after throwing an instance of 'std::runtime_error'
  what():  This executable was compiled without support for GpuVeryFastIntegralChannelsDetector
Aborted (core dumped)

Any Idea what's going on? And, more to the point, how can I fix this?

Thanks,

-Alex

Comments (38)

  1. fahall

    Huh... looks like I wasn't logged in when I posted before.

    I ran with OMP_NUM_THREADS=1 and it looks like I get the same error. For completeness, here is the console output:

    $ OMP_NUM_THREADS=1 ./objects_detection -c cvpr2012_very_fast_over_bahnhof.config.ini --gui.disable false
    Stereo objects detection on a flat world. Rodrigo Benenson, Markus Mathias, Mohamed Omran @ KULeuven & MPI-Inf. 2011-2014.
    Parsed the configuration file cvpr2012_very_fast_over_bahnhof.config.ini
    Using stereo camera calibration file: ../../video_input/calibration/stereo_calibration_bahnhof.proto.txt
    2015-03-14 23:05:08 {7f8b67c62ac0} [ StereoCameraCalibration ] : stereo_calibration_data name: Andreas Ess Bahnhofstrasse sequence stereo calibration
    Reading files:
    ../../../data/sample_test_images/bahnhof/image_00000000_0.png
    ../../../data/sample_test_images/bahnhof/image_00000000_1.png
    libprotobuf ERROR google/protobuf/message_lite.cc:123] Can't parse message of type "doppia_protobuf.DetectorModel" because it is missing required fields: detector_type
    2015-03-14 23:05:08 {7f8b67c62ac0} [ ObjectsDetectorFactory ] : Model file ../../../data/trained_models/2012_04_04_1417_trained_model_multiscales_synthetic_softcascade.proto.bin is not in binary protocol buffer format
    libprotobuf ERROR google/protobuf/text_format.cc:172] Error parsing text-format doppia_protobuf.DetectorModel: 2:1: Invalid control characters encountered in text.
    libprotobuf ERROR google/protobuf/text_format.cc:172] Error parsing text-format doppia_protobuf.DetectorModel: 2:9: Message type "doppia_protobuf.DetectorModel" has no field named "Scales".
    2015-03-14 23:05:08 {7f8b67c62ac0} [ ObjectsDetectorFactory ] : Model file ../../../data/trained_models/2012_04_04_1417_trained_model_multiscales_synthetic_softcascade.proto.bin is not in text protocol buffer format
    2015-03-14 23:05:08 {7f8b67c62ac0} [ ObjectsDetectorFactory ] : Parsed the binary model file ../../../data/trained_models/2012_04_04_1417_trained_model_multiscales_synthetic_softcascade.proto.bin
    2015-03-14 23:05:08 {7f8b67c62ac0} [ ObjectsDetectorFactory ] : Parsed the binary model file ../../../data/trained_models/2012_04_04_1417_trained_model_multiscales_synthetic_softcascade.proto.bin
    add_cascade_offset: at stage 0 the offset value is -0.05000
    add_cascade_offset: at stage 1000 the offset value is -0.05000
    add_cascade_offset: at stage 1999 the offset value is -0.05000
    add_cascade_offset: at stage 0 the offset value is -0.05000
    add_cascade_offset: at stage 1000 the offset value is -0.05000
    add_cascade_offset: at stage 1999 the offset value is -0.05000
    add_cascade_offset: at stage 0 the offset value is -0.05000
    add_cascade_offset: at stage 1000 the offset value is -0.05000
    add_cascade_offset: at stage 1999 the offset value is -0.05000
    add_cascade_offset: at stage 0 the offset value is -0.05000
    add_cascade_offset: at stage 1000 the offset value is -0.05000
    add_cascade_offset: at stage 1999 the offset value is -0.05000
    add_cascade_offset: at stage 0 the offset value is -0.05000
    add_cascade_offset: at stage 1000 the offset value is -0.05000
    add_cascade_offset: at stage 1999 the offset value is -0.05000
    2015-03-14 23:05:08 {7f8b67c62ac0} [ MultiScalesIntegralChannelsModel ] : Parsing multiscales model Scales 0.5, 1.0, 2.0, 4.0, 8.0
    2015-03-14 23:05:08 {7f8b67c62ac0} [ MultiScalesIntegralChannelsModel ] : Multiscales/bundle model is being normalized to have maximum detection score == 1
    2015-03-14 23:05:08 {7f8b67c62ac0} [ MultiScalesIntegralChannelsModel ] : Model index 0 has scale 0.5
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : Detector model is being normalized to have maximum detection score == 1
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : The soft cascade contains 2000 stages
    Stage 0 cascade_threshold == -0.010148
    Stage 999 cascade_threshold == -0.026132
    Stage 1999 cascade_threshold == -0.010174
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : Warning: No semantic category found in model, assuming a pedestrian detector
    2015-03-14 23:05:08 {7f8b67c62ac0} [ check_stages_and_range_visitor ] : Warning: Scale -1 is almost fine, min (x, y) == (0, 0), max (x,y) == (7, 15), width, height == (8, 16), (only touches left, top, border(s))
    2015-03-14 23:05:08 {7f8b67c62ac0} [ check_stages_and_range_visitor ] : Warning: Not touching the borders indicates that the model is suboptimal regarding detection quality (was trained with doppia v1, instead of v2)
    Single scale/occlusion model sanity check passed.
    2015-03-14 23:05:08 {7f8b67c62ac0} [ MultiScalesIntegralChannelsModel ] : Model index 1 has scale 1
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : Detector model is being normalized to have maximum detection score == 1
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : The soft cascade contains 2000 stages
    Stage 0 cascade_threshold == -0.010119
    Stage 999 cascade_threshold == -0.026103
    Stage 1999 cascade_threshold == -0.010146
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : Warning: No semantic category found in model, assuming a pedestrian detector
    2015-03-14 23:05:08 {7f8b67c62ac0} [ check_stages_and_range_visitor ] : Warning: Scale -1 is almost fine, min (x, y) == (0, 0), max (x,y) == (15, 31), width, height == (16, 32), (only touches left, top, border(s))
    Single scale/occlusion model sanity check passed.
    2015-03-14 23:05:08 {7f8b67c62ac0} [ MultiScalesIntegralChannelsModel ] : Model index 2 has scale 2
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : Detector model is being normalized to have maximum detection score == 1
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : The soft cascade contains 2000 stages
    Stage 0 cascade_threshold == -0.010106
    Stage 999 cascade_threshold == -0.026090
    Stage 1999 cascade_threshold == -0.010132
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : Warning: No semantic category found in model, assuming a pedestrian detector
    2015-03-14 23:05:08 {7f8b67c62ac0} [ check_stages_and_range_visitor ] : Warning: Scale -1 is almost fine, min (x, y) == (0, 0), max (x,y) == (31, 63), width, height == (32, 64), (only touches left, top, border(s))
    Single scale/occlusion model sanity check passed.
    2015-03-14 23:05:08 {7f8b67c62ac0} [ MultiScalesIntegralChannelsModel ] : Model index 3 has scale 4
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : Detector model is being normalized to have maximum detection score == 1
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : The soft cascade contains 2000 stages
    Stage 0 cascade_threshold == -0.010108
    Stage 999 cascade_threshold == -0.026092
    Stage 1999 cascade_threshold == -0.010135
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : Warning: No semantic category found in model, assuming a pedestrian detector
    2015-03-14 23:05:08 {7f8b67c62ac0} [ check_stages_and_range_visitor ] : Warning: Scale -1 is almost fine, min (x, y) == (0, 0), max (x,y) == (63, 127), width, height == (64, 128), (only touches left, top, border(s))
    Single scale/occlusion model sanity check passed.
    2015-03-14 23:05:08 {7f8b67c62ac0} [ MultiScalesIntegralChannelsModel ] : Model index 4 has scale 8
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : Detector model is being normalized to have maximum detection score == 1
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : The soft cascade contains 2000 stages
    Stage 0 cascade_threshold == -0.010115
    Stage 999 cascade_threshold == -0.026099
    Stage 1999 cascade_threshold == -0.010142
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : Warning: No semantic category found in model, assuming a pedestrian detector
    2015-03-14 23:05:08 {7f8b67c62ac0} [ check_stages_and_range_visitor ] : Warning: Scale -1 is almost fine, min (x, y) == (0, 0), max (x,y) == (127, 255), width, height == (128, 256), (only touches left, top, border(s))
    Single scale/occlusion model sanity check passed.
    2015-03-14 23:05:08 {7f8b67c62ac0} [ SoftCascadeOverIntegralChannelsModel ] : The detector model seems to include a cascade thresholds
    2015-03-14 23:05:08 {7f8b67c62ac0} [ ObjectsDetectorFactory ] : Updating the score threshold using the last cascade threshold of detector 0. New score threshold == -0.010 (= 0.000 + -0.010)
    A std::exception was raised: This executable was compiled without support for GpuVeryFastIntegralChannelsDetector
    terminate called after throwing an instance of 'std::runtime_error'
      what():  This executable was compiled without support for GpuVeryFastIntegralChannelsDetector
    Aborted (core dumped)
    
  2. Anoop Vargheese

    I got the same error. It's because the CMakeLists.txt file hasn't defined the USE_GPU constant, so it never compiled the Gpu code sections. It should be fixed if you add the line

    set(USE_GPU ON CACHE BOOL "Should the GPU be used ?" )
    

    to the beginning of the CMakeLists.txt file, and recompile.

  3. hu jianping

    Hello,Anonymous See your issue,I know you built object_detection successfully.But have you occured to the problem like that

    /usr/local/include/boost/variant/get.hpp:178:5: error: invalid application of sizeof to incomplete type boost::STATIC_ASSERTION_FAILURE<false>
         BOOST_STATIC_ASSERT_MSG(
         ^
    In file included from /home/hjp/doppia/src/applications/objects_detection/../../../libs/boost/geometry/geometries/point.hpp:26:0,
                     from /home/hjp/doppia/src/applications/objects_detection/../../../libs/boost/geometry/geometries/point_xy.hpp:22,
                     from /home/hjp/doppia/src/objects_detection/integral_channels/IntegralChannelsForPedestrians.hpp:13,
                     from /home/hjp/doppia/src/objects_detection/BaseVeryFastIntegralChannelsDetector.cpp:3:
    /home/hjp/doppia/src/applications/objects_detection/../../../libs/boost/geometry/util/math.hpp: At global scope:
    /home/hjp/doppia/src/applications/objects_detection/../../../libs/boost/geometry/util/math.hpp:176:14: warning: boost::geometry::math::r2d defined but not used [-Wunused-variable]
     double const r2d = 1.0 / d2r;
                  ^
    make[2]: *** [CMakeFiles/cpp_stuff_library.dir/home/hjp/doppia/src/objects_detection/BaseVeryFastIntegralChannelsDetector.cpp.o] Error 1
    make[1]: *** [CMakeFiles/cpp_stuff_library.dir/all] Error 2
    make[1]: *** Waiting ....
    

    If you have,could tell me how you solve it?Thank you advance!

  4. agert

    I have exactly the same problem. All the steps before worked fine and without problems. Similar to Isaac Robinson changing CMakeLists.txt did not help. Did this issue get solved yet?

  5. suresh

    even i faced the same "GpuVeryFastIntegralChannelsDetector error" issue while running in the CPU mode .can anyone help me with this one .Thank you

  6. Praveen Kumar

    I am facing "compiled without support for GpuVeryFastIntegralChannelsDetector" while running the object_detection.

    2016-03-03 16:16:37 {f403a000} [ check_stages_and_range_visitor ] : Warning: Scale -1 is almost fine, min (x, y) == (0, 0), max (x,y) == (127, 255), width, height == (128, 256), (only touches left, top, border(s))
    Single scale/occlusion model sanity check passed.
    2016-03-03 16:16:37 {f403a000} [ SoftCascadeOverIntegralChannelsModel ] : The detector model seems to include a cascade thresholds
    2016-03-03 16:16:37 {f403a000} [ ObjectsDetectorFactory ] : Updating the score threshold using the last cascade threshold of detector 0. New score threshold == -0.010 (= 0.000 + -0.010)
    A std::exception was raised: This executable was compiled without support for GpuVeryFastIntegralChannelsDetector
    terminate called after throwing an instance of 'std::runtime_error'
      what():  This executable was compiled without support for GpuVeryFastIntegralChannelsDetector
    Aborted
    
  7. suresh

    @praveen kumar you can get this issue resolved by changing line 38 as "method = cpu_very_fast" in "cvpr2012_very_fast_over_bahnhof.config.ini" in object_detection folder.But after this has been solved you may get program running successfully but may show the detected objects as 0.

    hope this is useful and please let me know if you could detect the objects . Thank you.

  8. Anoop Vargheese

    There's only one line in the entire code that generates this error: line 634 of the file doppia/src/objects_detection/ObjectsDetectorFactory.cpp. In other words, when it's trying to select the detection method at runtime.

    The summary of that function is essentially to check the value of the objects_detector.method variable in the config.ini file against one of these strings:

    • cpu_channel, cpu_channels or cpu_chnftrs : Throw a runtime error if a valid model has not been found, otherwise initialize the MultiscalesIntegralChannelsDetector classifier.
    • gpu_channel, gpu_channels or gpu_chnftrs : If the USE_GPU define has been set, then initialize the GpuMultiscalesIntegralChannelsDetector classifier. Otherwise throw a runtime error.
    • cpu_linear_svm, cpu_fpdw, gpu_fpdw : Throw error as this function has not yet been developed.
    • cpu_very_fast, cpu_fast, cpu_cvpr2012 : Throw an error if a valid model has not been found, otherwise initialize the VeryFastIntegralChannelsDetector classifier.
    • gpu_very_fast, gpu_fast or gpu_cvpr2012 : If the USE_GPU define has not been set, then throw a runtime error. Otherwise, initialize the GpuVeryFastIntegralChannelsDetector classifier.

    Your error falls under the last case, where all the config files are meant to use one of the GPU functions, but throws the runtime error because the USE_GPU define hasn't been set. Of course, you can bypass this error by using the cpu classifiers instead, as suggested by @suresh, but that doesn't solve the problem, as the whole point of the doppia library is to use the gpu functions.

    If my earlier suggestion of changing the cmake file didn't work, then I recommend changing the compile command to force it to be included, as shown below: cmake -D CMAKE_BUILD_TYPE=RelWithDebInfo -D USE_GPU=ON .

  9. suresh

    its interesting to know your valuable suggestion @Anoop Varghese .even i ll try tochk this option which might clear my 0 detection problem . anyway thank you for your valuable reply and i request you to that u can suggest any solutions more for my 0 detection problem

  10. Anoop Vargheese

    When I said "initialize the xxx classifier", I meant that was the logic of the program, i.e. those are the decisions that the code is making. I didn't imply that you had to do anything there.

    My only recommendation was to run the compile command with the cmake command : cmake -D CMAKE_BUILD_TYPE=RelWithDebInfo -D USE_GPU=ON . && make

    And just to make sure cmake has reconfigured the makefiles, I would recommend removing the following cmake files before running the above command :

    • CMakeCache.txt
    • CMakeFiles/
    • cmake_install.cmake
    • Makefile
  11. Anoop Vargheese

    If you don't want to do that, here's a link to the CMakeLists.txt file I used to compile. It's the same as the original cmake file from the doppia library, but I added two main changes:

    • On line 32, I set the USE_GPU flag as I'd mentioned before
    • On line 62, I specified the architecture of my GPU to compile for

    There are several other changes I made later in order to fit my compile, including upgrading the code to use the SDL2 libraries instead of SDL, but it should still compile without any problems. If you want the complete code that I forked and modified, you can find it on my github page here.

  12. Anoop Vargheese

    Ah, that's perfect. That means it's finally trying to compile the CUDA code, which it didn't even attempt earlier. Technically this fixes this issue, and you will no longer get the "compiled without support for GpuVeryFastIntegralChannelsDetector" error.

    Now you need to resolve this new compile error, which is caused by your CUDA libraries being incorrectly configured. I would start by looking at line 62 of my CMakeLists.txt file, and specify your correct CUDA architecture there. In my case, i made it compile for the cuda 2.0 architecture. Let me know if that fixes it? Can you also specify if you only copied the CMakeLists I shared, or if you're trying to compile my entire fork?

  13. suresh
                 from /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.hpp:5,
                 from /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp:1:
    

    /usr/include/c++/4.8/bits/shared_ptr_base.h:261:11: note: template<class Tp> class std::shared_ptr class shared_ptr; ^ In file included from /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.hpp:10:0, from /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp:1: /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/../../../src/applications/objects_detection/ObjectsDetectionApplication.hpp:46:5: error: ‘shared_ptr’ does not name a type shared_ptr<AbstractStixelWorldEstimator> stixel_world_estimator_p; ^ /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp: In constructor ‘doppia::ObjectsDetectionGui::ObjectsDetectionGui(doppia::ObjectsDetectionApplication&, const boost::program_options::variables_map&)’: /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp:127:49: error: ‘class doppia::ObjectsDetectionApplication’ has no member named ‘video_input_p’ : StixelWorldGui(application, application_.video_input_p, ^ /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp:128:35: error: ‘class doppia::ObjectsDetectionApplication’ has no member named ‘stixel_world_estimator_p’ application_.stixel_world_estimator_p, shared_ptr<AbstractStixelMotionEstimator>(), ^ /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp:149:24: error: ‘class doppia::ObjectsDetectionApplication’ has no member named ‘video_input_p’ if(application.video_input_p) ^ /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp:151:38: error: ‘class doppia::ObjectsDetectionApplication’ has no member named ‘video_input_p’ input_view = application.video_input_p->get_left_image(); ^ /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp:202:21: error: ‘class doppia::ObjectsDetectionApplication’ has no member named ‘stixel_world_estimator_p’ if(application_.stixel_world_estimator_p) ^ /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp: In member function ‘virtual bool doppia::ObjectsDetectionGui::process_inputs()’: /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp:330:89: error: ‘class doppia::ObjectsDetectionApplication’ has no member named ‘video_input_p’ VideoFromFiles video_from_files_p = dynamic_cast<VideoFromFiles >(application.video_input_p.get()); ^ /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp: In member function ‘void doppia::ObjectsDetectionGui::resize_if_necessary()’: /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp:348:20: error: ‘class doppia::ObjectsDetectionApplication’ has no member named ‘video_input_p’ if(application.video_input_p) ^ /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp: In member function ‘void doppia::ObjectsDetectionGui::draw_video_input()’: /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp:382:20: error: ‘class doppia::ObjectsDetectionApplication’ has no member named ‘video_input_p’ if(application.video_input_p) ^ /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp:384:85: error: ‘class doppia::ObjectsDetectionApplication’ has no member named ‘video_input_p’ const AbstractVideoInput::input_image_view_t &left_input_view = application.video_input_p->get_left_image(); ^ /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp:385:86: error: ‘class doppia::ObjectsDetectionApplication’ has no member named ‘video_input_p’ const AbstractVideoInput::input_image_view_t &right_input_view = application.video_input_p->get_right_image(); ^ /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp: In member function ‘void doppia::ObjectsDetectionGui::draw_tracks()’: /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp:543:48: error: ‘class doppia::ObjectsDetectionApplication’ has no member named ‘video_input_p’ &left_input_view = application.video_input_p->get_left_image(); ^ /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/ObjectsDetectionGui.cpp:604:21: error: ‘stixel_world_estimator_p’ was not declared in this scope stixel_world_estimator_p.get(),

  14. suresh

    i had changed my cmakelist.txt line 251 to set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11") # required for unrestricted unions this solved the previous problem of float and gave this error

  15. suresh

    /usr/local/include/boost/config/suffix.hpp(510): error: identifier "__float128" is undefined

    /usr/local/include/boost/type_traits/is_floating_point.hpp(25): error: identifier "__float128" is undefined

    /usr/local/include/boost/type_traits/type_with_alignment.hpp(48): error: identifier "__float128" is undefined

    3 errors detected in the compilation of "/tmp/tmpxft_0000127f_00000000-6_integral_channels_detector.cpp1.ii". CMake Error at cuda_stuff_library_generated_integral_channels_detector.cu.o.cmake:264 (message): Error generating file /home/hubino/Downloads/pdt_360deg_git-master/src/applications/objects_detection/CMakeFiles/cuda_stuff_library.dir///__/src/objects_detection/gpu/./cuda_stuff_library_generated_integral_channels_detector.cu.o

  16. Anoop Vargheese

    Well, to tell the truth, I was using Boost v1.54. I haven't tried using this with more recent versions of the library, but I'm fairly certain this isn't a problem with boost, even though it says the error was found in the boost headers. It still looks like an error with CUDA.

    Just to verify, did you set the correct CUDA architecture in the CMakeLists.txt file? If so, which one?

  17. suresh

    i m using cuda 6.5 which i had replaced everywhere need to be changed.is this what u mean by architecture.if not can u suggest me how to know it

  18. Anoop Vargheese

    Ah no, that's not supposed to be the cuda version. You need to set the compute-capability of your cuda card. You can find the list of compute-capabilities of each card on this page : https://developer.nvidia.com/cuda-gpus. For example, I used the command set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS} -arch compute_20 -code sm_20) because I was using the NVIDIA GeForce 540M, and I was compiling for the 2.0 architecture.
    An alternate way to find the architecture of your cuda card is to compile the deviceQuery.cu code (either present in your CUDA samples directory, or else you can compile this source directly using nvcc. The Major and Minor revision numbers together form the compute-capability or architecture of the card to compile for.

  19. suresh

    Downgrading to boost 1.57.0 solved the issue but finally got up with one more error

    Linking CXX static library libcuda_stuff_library.a [ 99%] Built target cuda_stuff_library make[2]: No rule to make target /usr/lib/nvidia/libcuda.so', needed byobjects_detection'. Stop. make[1]: [CMakeFiles/objects_detection.dir/all] Error 2 make: *** [all] Error 2

  20. suresh

    Device 0: "GeForce GT 640" CUDA Driver Version / Runtime Version 7.5 / 6.5 CUDA Capability Major/Minor version number: 3.0

    it means set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS} -arch compute_30 -code sm_30) like this or not?

  21. suresh

    Thanx a lot anoop and also every one who posted many useful answers to this group ......... now i could successfully run object detection and detect the pedestrians and their faces

  22. suresh

    Hi Liuwei................. I suggest you to follow up the above discussion which would solve all the major issues..... and for the issue you are facing i think u could find a solution if u go through the other discussions also....... It was almost more than a year i worked on it so i suggest you to follow up here......

  23. Log in to comment