Skip to content

Latest commit

 

History

History
328 lines (251 loc) · 9.93 KB

README_EN.md

File metadata and controls

328 lines (251 loc) · 9.93 KB

examples

中文版本

Tengine Lite's examples providing simple yet fancy demos.

Classification task - tm_classification.c

Tengine Lite is compatible with original Tengine's C API. Here we demonstrate how to run MobileNet v1 via tm_classification example code, providing image classification functionality. This would help you get involve with Tengine Lite C API. We use the popular tiger cat image for test.

lu mao

Model files can be found here: Tengine model zoo, which are compatible with original Tengine example models（password：hhgc）。

Reference Code

tm_classification.c

Compilation

build.sh compiles example folders demo programs on default. Take x86 platform as example, the compilation generated demos are stored in ./build/install/bin/ folder.

bug1989@DESKTOP-SGN0H2A:/mnt/d/ubuntu/gitlab/build-linux$ tree install
install
├── bin
│   ├── cpp_tm_classification
│   ├── cpp_tm_mobilenet_ssd
│   ├── tm_benchmark
│   ├── tm_classification
│   ├── tm_classification_fp16
│   ├── tm_classification_uint8
│   ├── tm_classification_vulkan
│   ├── tm_crnn
│   ├── tm_landmark
│   ├── tm_landmark_uint8
│   ├── tm_mobilefacenet
│   ├── tm_mobilenet_ssd
│   ├── tm_mobilenet_ssd_acl
│   ├── tm_mobilenet_ssd_uint8
│   ├── tm_openpose
│   ├── tm_retinaface
│   ├── tm_yolact
│   ├── tm_yolov3_tiny
│   ├── tm_yolov3_uint8
│   ├── tm_yolov4
│   └── tm_yolov4_tiny
├── include
│   └── tengine_c_api.h
└── lib
    └── libtengine-lite.so

Running Result

Put testing images and models under root folder of Tengine-Lite project, and run it:

export LD_LIBRARY_PATH=./build/install/lib
./build/install/bin/tm_classification -m models/mobilenet.tmfile -i images/cat.jpg -g 224,224 -s 0.017,0.017,0.017 -w 104.007,116.669,122.679

output:

start to run register cpu allocator
tengine-lite library version: 1.0-dev

model file : ./temp/models/mobilenet.tmfile
image file : ./temp/images/cat.jpg
img_h, img_w, scale[3], mean[3] : 224 224 , 0.017 0.017 0.017, 104.0 116.7 122.7
Repeat 1 times, thread 1, avg time 656.76 ms, max_time 656.76 ms, min_time 656.76 ms
--------------------------------------
8.574148, 282
7.880116, 277
7.812579, 278
7.286453, 263
6.357488, 281
--------------------------------------

Facial Landmark Detection Task - tm_landmark.cpp

We use this image:

Run it with:

export LD_LIBRARY_PATH=./build/install/lib
./build/install/bin/tm_landmark -m models/landmark.tmfile -i images/mobileface02.jpg -r 1 -t 1

output:

start to run register cpu allocator
tengine-lite library version: 1.0-dev
Repeat [1] min 17.461 ms, max 17.461 ms, avg 17.461 ms

SSD Object Detection Task - tm_mobilenet_ssd.cpp

We use this image:

export LD_LIBRARY_PATH=./build/install/lib
./build/install/bin/tm_mobilenet_ssd -m models/mobilenet_ssd.tmfile -i images/ssd_dog.jpg -r 1 -t 1

output:

start to run register cpu allocator
tengine-lite library version: 1.0-dev
Repeat 1 times, thread 1, avg time 206.30 ms, max_time 206.30 ms, min_time 206.30 ms
--------------------------------------
detect result num: 3
dog	:99.8%
BOX:( 138 , 209 ),( 324 , 541 )
car	:99.7%
BOX:( 467 , 72 ),( 687 , 171 )
bicycle	:99.5%
BOX:( 107 , 141 ),( 574 , 415 )
======================================
[DETECTED IMAGE SAVED]:
======================================

RetinaFace Face Detection Task - tm_refinaface.cpp

We use this image:

export LD_LIBRARY_PATH=./build/install/lib
./build/install/bin/tm_retinaface -m models/retinaface.tmfile -i images/mtcnn_face4.jpg -r 1 -t 1

output：

start to run register cpu allocator
tengine-lite library version: 1.0-dev
img_h, img_w : 316, 474
Repeat 1 times, thread 1, avg time 75.72 ms, max_time 75.72 ms, min_time 75.72 ms
--------------------------------------
detected face num: 4
BOX 0.99:( 38.9179 , 86.3346 ),( 45.7028 , 63.2934 )
BOX 0.99:( 168.12 , 86.14 ),( 37.5249 , 47.7839 )
BOX 0.98:( 383.673 , 56.4136 ),( 77.176 , 83.8093 )
BOX 0.98:( 289.365 , 103.773 ),( 38.0025 , 47.6989 )

Yolact Instance Segmentation Task - tm_yolact.cpp

We use this image:

export LD_LIBRARY_PATH=./build/install/lib
./build/install/bin/tm_yolact -m models/yolact.tmfile -i images/ssd_car.jpg -r 1 -t 1

output:

start to run register cpu allocator
tengine-lite library version: 1.0-dev
Repeat 1 times, thread 1, avg time 15833.47 ms, max_time 15833.47 ms, min_time 15833.47 ms
--------------------------------------
6 = 0.99966 at 130.82 57.77 340.78 x 237.36
3 = 0.99675 at 323.39 194.97 175.57 x 132.96
1 = 0.33431 at 191.24 195.78 103.06 x 179.22

YoloV3 Object Detection Task - tm_yolov3.cpp

We use this image:

export LD_LIBRARY_PATH=./build/install/lib
./build/install/bin/tm_yolact -m models/yolov3_tiny.tmfile -i images/ssd_dog.jpg -r 1 -t 1

output：

start to run register cpu allocator
tengine-lite library version: 1.0-dev
Repeat 1 times, thread 1, avg time 262.52 ms, max_time 262.52 ms, min_time 262.52 ms
--------------------------------------
num_detections,4
16: 57%
left = 129,right = 369,top = 186,bot = 516
2: 65%
left = 465,right = 677,top = 74,bot = 171
1: 60%
left = 205,right = 576,top = 153,bot = 447

Yolov4-tiny Object Detection Task - tm_yolov4_tiny.cpp

We use this image:

$ export LD_LIBRARY_PATH=./build/install/lib
$ ./build/install/bin/tm_yolov4_tiny -m models/yolov4_tiny.tmfile -i images/ssd_dog.jpg -r 1 -t 1

output：

start to run register cpu allocator
tengine-lite library version: 1.0-dev
Repeat 1 times, thread 1, avg time 177.72 ms, max_time 177.72 ms, min_time 177.72 ms
--------------------------------------
num_detections,10
16: 74%
left = 125,right = 327,top = 221,bot = 537
2: 40%
7: 84%
left = 455,right = 703,top = 77,bot = 168
1: 28%
left = 56,right = 603,top = 85,bot = 496

Human Pose Estimation Task - tm_openpose.cpp

We use this image:

export LD_LIBRARY_PATH=./build/install/lib
./build/install/bin/tm_openpose -m models/openpose_coco.tmfile -i image/pose.jpg -r 1 -t 1

output:

start to run register cpu allocator
tengine-lite library version: 1.0-dev
Repeat 1 times, thread 1, avg time 15350.25 ms, max_time 15350.25 ms, min_time 15350.25 ms
--------------------------------------
KeyPoints Coordinate:
0:[292.174, 55.6522]
1:[306.087, 125.217]
2:[250.435, 139.13]
3:[236.522, 222.609]
4:[222.609, 306.087]
5:[361.739, 125.217]
6:[403.478, 208.696]
7:[417.391, 292.174]
8:[264.348, 306.087]
9:[264.348, 431.304]
10:[264.348, 570.435]
11:[347.826, 306.087]
12:[375.652, 431.304]
13:[333.913, 542.609]
14:[278.261, 41.7391]
15:[306.087, 41.7391]
16:[264.348, 55.6522]
17:[320, 55.6522]

The result of human pose estimation will be saved as images, whose names are: Output-Keypionts.jpg and Output-Skeleton.jpg.

Chinese character recognition task - tm_crnn.cpp

Folder structure: create model directory under project root directory, and download corresponding model and testing image files from Tengine model zoo ( password：hhgc ) there. Specifically, crnn_lite_dense.tmfile as model file, o2_resize.jpg as testing image, keys.txt as font file. Then we execute:

export LD_LIBRARY_PATH=./build/install/lib
./build/install/bin/tm_crnn -m model/crnn_lite_dense.tmfile -i model/o2_resize.jpg -l model/keys.txt

result:

start to run register cpu allocator
tengine-lite library version: 1.0-dev
Repeat 1 times, thread 1, avg time 43.32 ms, max_time 43.32 ms, min_time 43.32 ms
--------------------------------------
如何突破自己的颜值上限
--------------------------------------

The result of ocr recognition is displayed in terminal, you may also modify the source code to save it to file.

We will continously updating more fancy demos, please stay tuned...