search

enazoe / yolo-tensorrt

TensorRT8.Support Yolov5n,s,m,l,x .darknet -> tensorrt. Yolov4 Yolov3 use raw darknet *.weights and *.cfg fils. If the wrapper is useful to you,please Star it.
MIT License
1.19k stars 315 forks source link

yolov3-tiny #27

Closed Jessicahjj closed 4 years ago

Jessicahjj commented 4 years ago

Config config_v3; config_v3.net_type = YOLOV3_TINY; config_v3.file_model_cfg = "../configs/yolov3-tiny.cfg"; config_v3.file_model_weights = "../configs/yolov3-tiny.weights"; config_v3.calibration_image_list_file_txt = "../configs/calibration_images.txt"; config_v3.inference_precison = INT8;

yolov3, yolov4 and yolov4-tiny run successfully. but yolov3-tiny failed.

Loading pre-trained weights... Loading complete! layer inp_size out_size weightPtr (1) conv-bn-leaky 3 x 416 x 416 16 x 416 x 416 496 (2) maxpool 16 x 416 x 416 16 x 208 x 208 496 (3) conv-bn-leaky 16 x 208 x 208 32 x 208 x 208 5232 (4) maxpool 32 x 208 x 208 32 x 104 x 104 5232 (5) conv-bn-leaky 32 x 104 x 104 64 x 104 x 104 23920 (6) maxpool 64 x 104 x 104 64 x 52 x 52 23920 (7) conv-bn-leaky 64 x 52 x 52 128 x 52 x 52 98160 (8) maxpool 128 x 52 x 52 128 x 26 x 26 98160 (9) conv-bn-leaky 128 x 26 x 26 256 x 26 x 26 394096 (10) maxpool 256 x 26 x 26 256 x 13 x 13 394096 (11) conv-bn-leaky 256 x 13 x 13 512 x 13 x 13 1575792 (12) maxpool 512 x 13 x 13 512 x 12 x 12 1575792 (13) conv-bn-leaky 512 x 12 x 12 1024 x 12 x 12 6298480 (14) conv-bn-leaky 1024 x 12 x 12 256 x 12 x 12 6561648 (15) conv-bn-leaky 256 x 12 x 12 512 x 12 x 12 7743344 (16) conv-linear 512 x 12 x 12 255 x 12 x 12 7874159 (17) yolo 255 x 12 x 12 255 x 12 x 12 7874159 (18) route - 256 x 12 x 12 7874159 (19) conv-bn-leaky 256 x 12 x 12 128 x 12 x 12 7907439 (20) upsample 128 x 12 x 12 128 x 24 x 24 - ERROR: Assertion failed: d.nbDims >= 1, file f:\image-process-lib\code\tensorrt-7.0.0.11\samples\yolo-tensorrt\modules\trt_utils.cpp, line 409

enazoe commented 4 years ago

@JiaojiaoHan is a bug，i will fix it soon.

beizhengren commented 4 years ago

yolov3, yolov4 and yolov4-tiny run successfully.

@enazoe @JiaojiaoHan Hi, there are some errors when I compling the repo. Have you encountered that ? How do you resolve it? Thank you!

/home/nvidia/Documents/wyz/darknetTensorRT/yolo-tensorrt/modules/chunk.h(35): error: cannot override "final" function "nvinfer1::IPluginV2IOExt::supportsFormat"
/usr/include/aarch64-linux-gnu/NvInferRuntimeCommon.h(762): here

/home/nvidia/Documents/wyz/darknetTensorRT/yolo-tensorrt/modules/chunk.h(46): warning: function "nvinfer1::IPluginV2Ext::configurePlugin(const nvinfer1::Dims *, int, const nvinfer1::Dims *, int, const nvinfer1::DataType *, const nvinfer1::DataType *, const __nv_bool *, const __nv_bool *, nvinfer1::PluginFormat, int)" is hidden by "nvinfer1::Chunk::configurePlugin" -- virtual function override intended?

/home/nvidia/Documents/wyz/darknetTensorRT/yolo-tensorrt/modules/chunk.cu(70): error: identifier "rsize_t" is undefined

/home/nvidia/Documents/wyz/darknetTensorRT/yolo-tensorrt/modules/chunk.cu(70): error: declaration is incompatible with "size_t nvinfer1::Chunk::getSerializationSize() const"
/home/nvidia/Documents/wyz/darknetTensorRT/yolo-tensorrt/modules/chunk.h(33): here

3 errors detected in the compilation of "/tmp/tmpxft_000045c1_00000000-6_chunk.cpp1.ii".
CMake Error at detector_generated_chunk.cu.o.cmake:280 (message):
  Error generating file
  /home/nvidia/Documents/wyz/darknetTensorRT/yolo-tensorrt/build-nx/CMakeFiles/detector.dir/modules/./detector_generated_chunk.cu.o

CMakeFiles/detector.dir/build.make:334: recipe for target 'CMakeFiles/detector.dir/modules/detector_generated_chunk.cu.o' failed
make[2]: *** [CMakeFiles/detector.dir/modules/detector_generated_chunk.cu.o] Error 1
CMakeFiles/Makefile2:124: recipe for target 'CMakeFiles/detector.dir/all' failed
make[1]: *** [CMakeFiles/detector.dir/all] Error 2
Makefile:103: recipe for target 'all' failed
make: *** [all] Error 2
enazoe commented 4 years ago

@beizhengren which version of jetpack?

beizhengren commented 4 years ago

@enazoe Hi, I used jetson xavier NX with tensorrt 7.1.0-1+cuda10.2
I have found some information by 

 cat /etc/nv_tegra_release

# R32 (release), REVISION: 4.2, GCID: 20074772, BOARD: t186ref, EABI: aarch64, DATE: Thu Apr 9 01:26:40 UTC 2020 And I can complie successfully with dev-yolov4 branch which not support yolov4-tiny.

enazoe commented 4 years ago

@beizhengren you could switch to master, it support yolov4-tiny now

beizhengren commented 4 years ago

you could switch to master, it support yolov4-tiny now

@enazoe Hi, I just used master branch, and encountered the errors https://github.com/enazoe/yolo-tensorrt/issues/27#issuecomment-661752953. I mean, master can not complie successfully, but dev-yolov4 can.

enazoe commented 4 years ago

@beizhengren i will fix it soon

beizhengren commented 4 years ago

@enazoe Hi I've resolved it, and run successfully! Thanks!

enazoe commented 4 years ago

@Jessicahjj try it 7898f6538a0a002a117465d62af74434d8f5e925

Jessicahjj commented 4 years ago

@enazoe Thanks!

myboyhood commented 1 year ago

Yes! I also meet this problem, and solve it.

  1. I checkout master
  2. recompile with cmake .. && make
  3. delete the old yolov3-tiny-kun-kHALF-batch1.engine file, because this file is generate according to the machine and GPU type. Let us to generate a new one.
  4. The successful ouput is as following: 
    darknet_weights_path: /home/zph/vins_ws/src/pnp_target_node/config/yolov3-tiny-kun.weights
darknet_cfg: /home/zph/vins_ws/src/pnp_target_node/config/yolov3-tiny-kun.cfgFile does not exist : /home/zph/vins_ws/src/pnp_target_node/config/yolov3-tiny-kun-kHALF-batch1.engine
Loading pre-trained weights...
Loading complete!
  layer               inp_size            out_size       weightPtr
(1)   conv-bn-leaky     1 x 416 x 416      16 x 416 x 416    208   
(2)   maxpool          16 x 416 x 416      16 x 208 x 208    208   
(3)   conv-bn-leaky    16 x 208 x 208      32 x 208 x 208    4944  
(4)   maxpool          32 x 208 x 208      32 x 104 x 104    4944  
(5)   conv-bn-leaky    32 x 104 x 104      64 x 104 x 104    23632 
(6)   maxpool          64 x 104 x 104      64 x  52 x  52    23632 
(7)   conv-bn-leaky    64 x  52 x  52     128 x  52 x  52    97872 
(8)   maxpool         128 x  52 x  52     128 x  26 x  26    97872 
(9)   conv-bn-leaky   128 x  26 x  26     256 x  26 x  26    393808
(10)  maxpool         256 x  26 x  26     256 x  13 x  13    393808
(11)  conv-bn-leaky   256 x  13 x  13     512 x  13 x  13    1575504
(12)  maxpool         512 x  13 x  13     512 x  13 x  13    1575504
(13)  conv-bn-leaky   512 x  13 x  13    1024 x  13 x  13    6298192
(14)  conv-bn-leaky  1024 x  13 x  13     256 x  13 x  13    6561360
(15)  conv-bn-leaky   256 x  13 x  13     512 x  13 x  13    7743056
(16)  conv-linear     512 x  13 x  13      18 x  13 x  13    7752290
(17)  yolo             18 x  13 x  13      18 x  13 x  13    7752290
(18)  route                  -            256 x  13 x  13    7752290
(19)  conv-bn-leaky   256 x  13 x  13     128 x  13 x  13    7785570
(20)  upsample        128 x  13 x  13     128 x  26 x  26        - 
(21)  route                  -            384 x  26 x  26    7785570
(22)  conv-bn-leaky   384 x  26 x  26     256 x  26 x  26    8671330
(23)  conv-linear     256 x  26 x  26      18 x  26 x  26    8675956
(24)  yolo             18 x  26 x  26      18 x  26 x  26    8675956
File does not exist : /home/zph/vins_ws/src/pnp_target_node/config/yolov3-tiny-kun-kHALF-batch1.engine
Building the TensorRT Engine...
Building complete!
Serializing the TensorRT Engine...
Serialized plan file cached at location : /home/zph/vins_ws/src/pnp_target_node/config/yolov3-tiny-kun-kHALF-batch1.engine
Loading TRT Engine...
Loading Complete!
TensorRTDetector initialized