format

.pre-commit-config.yaml

@@ -11,17 +11,17 @@ repos:
       - id: name-tests-test
       - id: requirements-txt-fixer
 
-  - repo: https://github.com/psf/black
-    rev: 21.6b0
-    hooks:
-      - id: black
-
   - repo: https://github.com/pycqa/isort
     rev: 5.8.0
     hooks:
       - id: isort
         args: ["--profile", "black"]
 
+  - repo: https://github.com/psf/black
+    rev: 21.6b0
+    hooks:
+      - id: black
+
   - repo: https://gitlab.com/pycqa/flake8
     rev: 3.9.2
     hooks:

LICENSE

@@ -1,5 +1,5 @@
  Copyright 2020-2021 RangiLyu. All rights reserved.
- 
+
                                 Apache License
                            Version 2.0, January 2004
                         http://www.apache.org/licenses/

README.md

@@ -44,8 +44,8 @@ Model          |Resolution|COCO mAP |Latency(ARM 4 Threads) | FLOPS      |   Par
 :-------------:|:--------:|:-------:|:--------------------:|:----------:|:---------:|:-------:
 NanoDet-m      | 320*320 |   20.6   | **10.23ms**          | **0.72G**  | **0.95M** | **1.8MB(FP16)** | **980KB(INT8)**
 NanoDet-m      | 416*416 |   23.5   | 16.44ms              | 1.2G       | **0.95M** | **1.8MB(FP16)** | **980KB(INT8)**
-NanoDet-m-1.5x | 320*320 |   23.5   | 13.53ms              | 1.44G      | 2.08M     |   3.9MB(FP16) | 2MB(INT8) 
-NanoDet-m-1.5x | 416*416 | **26.8** | 21.53ms              | 2.42G      | 2.08M     |   3.9MB(FP16) | 2MB(INT8) 
+NanoDet-m-1.5x | 320*320 |   23.5   | 13.53ms              | 1.44G      | 2.08M     |   3.9MB(FP16) | 2MB(INT8)
+NanoDet-m-1.5x | 416*416 | **26.8** | 21.53ms              | 2.42G      | 2.08M     |   3.9MB(FP16) | 2MB(INT8)
 NanoDet-g      | 416*416 |   22.9   | Not Designed For ARM | 4.2G       | 3.81M     |   7.7MB(FP16) | 3.6MB(INT8)
 YoloV3-Tiny    | 416*416 |   16.6   | 37.6ms               | 5.62G      | 8.86M     |   33.7MB
 YoloV4-Tiny    | 416*416 |   21.7   | 32.81ms              | 6.96G      | 6.06M     |   23.0MB
@@ -222,15 +222,15 @@ NanoDet-RepVGG        | RepVGG-A0          | 416*416  |  27.8  | 11.3G | 6.75M |
 3. **Start training**
 
    NanoDet is now using [pytorch lightning](https://github.com/PyTorchLightning/pytorch-lightning) for training.
-   
+
    For both single-GPU or multiple-GPUs, run:
-   
+
    ```shell script
    python tools/train.py CONFIG_FILE_PATH
    ```
-   
+
    For **Windows users**, if you have problems with the new lightning trainer, try to use tools/deprecated/train.py
-   
+
    <details>
    <summary>follow this...</summary>
 
@@ -245,7 +245,7 @@ NanoDet-RepVGG        | RepVGG-A0          | 416*416  |  27.8  | 11.3G | 6.75M |
     ```shell script
     python -m torch.distributed.launch --nproc_per_node=GPU_NUM --master_port 29501 tools/deprecated/train.py CONFIG_FILE_PATH
     ```
-   
+
    </details>
 
 

demo/demo-inference-with-pytorch.ipynb

@@ -260,4 +260,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 4
-}
+}

demo_android_ncnn/LICENSE

@@ -671,4 +671,4 @@ into proprietary programs.  If your program is a subroutine library, you
 may consider it more useful to permit linking proprietary applications with
 the library.  If this is what you want to do, use the GNU Lesser General
 Public License instead of this License.  But first, please read
-<http://www.gnu.org/philosophy/why-not-lgpl.html>.
+<http://www.gnu.org/philosophy/why-not-lgpl.html>.

demo_android_ncnn/README.md

@@ -5,7 +5,7 @@ This repo is an Android object detection demo of NanoDet using
 
 # Tutorial
 
-## Step1. 
+## Step1.
 Download ncnn-android-vulkan.zip from ncnn repo or build ncnn-android from source.
 
 - [ncnn-android-vulkan.zip download link](https://github.com/Tencent/ncnn/releases)
@@ -32,5 +32,4 @@ Open demo_android_ncnn folder with Android Studio and then build it.
 # Reference
 
 * [ncnn](https://github.com/tencent/ncnn)
-* [YOLOv5_NCNN](https://github.com/WZTENG/YOLOv5_NCNN) 
-
+* [YOLOv5_NCNN](https://github.com/WZTENG/YOLOv5_NCNN)

demo_android_ncnn/app/src/main/AndroidManifest.xml

@@ -27,4 +27,4 @@
             android:theme="@style/AppNoTitleTheme" />
     </application>
 
-</manifest>
+</manifest>

demo_android_ncnn/app/src/main/cpp/YoloV4.cpp

@@ -69,4 +69,3 @@ YoloV4::decode_infer(ncnn::Mat &data, const yolocv::YoloSize &frame_size, int ne
     }
     return result;
 }
-

demo_android_ncnn/app/src/main/cpp/jni_interface.cpp

@@ -107,6 +107,3 @@ Java_com_rangi_nanodet_YOLOv4_detect(JNIEnv *env, jclass, jobject image, jdouble
     }
     return ret;
 }
-
-
-

demo_android_ncnn/app/src/main/res/layout/activity_main.xml

@@ -107,4 +107,4 @@
         app:layout_constraintTop_toBottomOf="@+id/imageView"
         />
 
-</androidx.constraintlayout.widget.ConstraintLayout>
+</androidx.constraintlayout.widget.ConstraintLayout>

demo_android_ncnn/app/src/main/res/mipmap-anydpi-v26/ic_launcher.xml

@@ -2,4 +2,4 @@
 <adaptive-icon xmlns:android="http://schemas.android.com/apk/res/android">
     <background android:drawable="@drawable/ic_launcher_background" />
     <foreground android:drawable="@drawable/ic_launcher_foreground" />
-</adaptive-icon>
+</adaptive-icon>

demo_android_ncnn/app/src/main/res/mipmap-anydpi-v26/ic_launcher_round.xml

@@ -2,4 +2,4 @@
 <adaptive-icon xmlns:android="http://schemas.android.com/apk/res/android">
     <background android:drawable="@drawable/ic_launcher_background" />
     <foreground android:drawable="@drawable/ic_launcher_foreground" />
-</adaptive-icon>
+</adaptive-icon>

demo_android_ncnn/app/src/test/java/com/rangi/nanodet/ExampleUnitTest.java

@@ -14,4 +14,4 @@ public class ExampleUnitTest {
     public void addition_isCorrect() {
         assertEquals(4, 2 + 2);
     }
-}
+}

demo_android_ncnn/build.gradle

@@ -4,11 +4,11 @@ buildscript {
     repositories {
         google()
         jcenter()
-        
+
     }
     dependencies {
         classpath 'com.android.tools.build:gradle:4.0.0'
-        
+
         // NOTE: Do not place your application dependencies here; they belong
         // in the individual module build.gradle files
     }

demo_android_ncnn/gradle.properties

@@ -17,4 +17,3 @@ org.gradle.jvmargs=-Xmx1536m
 android.useAndroidX=true
 # Automatically convert third-party libraries to use AndroidX
 android.enableJetifier=true
-

demo_libtorch/CMakeLists.txt

@@ -16,9 +16,9 @@ include_directories(
 add_executable(nanodet_demo main.cpp nanodet_libtorch.cpp)
 
 target_link_libraries(
-    nanodet_demo 
+    nanodet_demo
     ${TORCH_LIBRARIES}
-    ${OpenCV_LIBS} 
+    ${OpenCV_LIBS}
 )
 
 if (MSVC)
@@ -28,4 +28,4 @@ if (MSVC)
                      COMMAND ${CMAKE_COMMAND} -E copy_if_different
                      ${TORCH_DLLS}
                      $<TARGET_FILE_DIR:nanodet_demo>)
-endif (MSVC)
+endif (MSVC)

demo_libtorch/README.md

@@ -1,6 +1,6 @@
 # NanoDet TorchScript / LibTorch Demo
 
-This folder provides NanoDet inference code using for LibTorch.  
+This folder provides NanoDet inference code using for LibTorch.
 
 ## Install dependencies
 
@@ -29,4 +29,4 @@ mkdir build
 cd build
 cmake -DCMAKE_PREFIX_PATH=/absolute/path/to/libtorch ..
 make
-```
+```

demo_libtorch/main.cpp

@@ -194,7 +194,7 @@ void draw_bboxes(const cv::Mat& bgr, const std::vector<BoxInfo>& bboxes, object_
         //fprintf(stderr, "%d = %.5f at %.2f %.2f %.2f %.2f\n", bbox.label, bbox.score,
         //    bbox.x1, bbox.y1, bbox.x2, bbox.y2);
 
-        cv::rectangle(image, cv::Rect(cv::Point((bbox.x1 - effect_roi.x) * width_ratio, (bbox.y1 - effect_roi.y) * height_ratio), 
+        cv::rectangle(image, cv::Rect(cv::Point((bbox.x1 - effect_roi.x) * width_ratio, (bbox.y1 - effect_roi.y) * height_ratio),
                                       cv::Point((bbox.x2 - effect_roi.x) * width_ratio, (bbox.y2 - effect_roi.y) * height_ratio)), color);
 
         char text[256];
@@ -292,15 +292,15 @@ int benchmark(NanoDet& detector)
     double time_max = -DBL_MAX;
     double time_avg = 0;
     cv::Mat image(320, 320, CV_8UC3, cv::Scalar(1, 1, 1));
-    
+
     for (int i = 0; i < warm_up + loop_num; i++)
     {
         auto start = std::chrono::steady_clock::now();
         std::vector<BoxInfo> results;
         results = detector.detect(image, 0.4, 0.5);
         auto end = std::chrono::steady_clock::now();
         double time = std::chrono::duration<double, std::milli>(end - start).count();
-        if (i >= warm_up) 
+        if (i >= warm_up)
         {
             time_min = (std::min)(time_min, time);
             time_max = (std::max)(time_max, time);
@@ -312,7 +312,7 @@ int benchmark(NanoDet& detector)
     return 0;
 }
 
-int main(int argc, char** argv) 
+int main(int argc, char** argv)
 {
     if (argc != 3)
     {

demo_libtorch/nanodet_libtorch.cpp

@@ -3,7 +3,7 @@
 #include "nanodet_libtorch.h"
 
 
-inline float fast_exp(float x) 
+inline float fast_exp(float x)
 {
     union {
         uint32_t i;
@@ -13,7 +13,7 @@ inline float fast_exp(float x)
     return v.f;
 }
 
-inline float sigmoid(float x) 
+inline float sigmoid(float x)
 {
     return 1.0f / (1.0f + fast_exp(-x));
 }
@@ -24,13 +24,13 @@ int activation_function_softmax(const _Tp* src, _Tp* dst, int length)
     const _Tp alpha = *std::max_element(src, src + length);
     _Tp denominator{ 0 };
 
-    for (int i = 0; i < length; ++i) 
+    for (int i = 0; i < length; ++i)
     {
         dst[i] = fast_exp(src[i] - alpha);
         denominator += dst[i];
     }
 
-    for (int i = 0; i < length; ++i) 
+    for (int i = 0; i < length; ++i)
     {
         dst[i] /= denominator;
     }
@@ -85,7 +85,7 @@ std::vector<BoxInfo> NanoDet::detect(cv::Mat image, float score_threshold, float
     for (int i = 0; i < (int)results.size(); i++)
     {
         this->nms(results[i], nms_threshold);
-        
+
         for (auto box : results[i])
         {
             dets.push_back(box);
@@ -161,14 +161,14 @@ void NanoDet::nms(std::vector<BoxInfo>& input_boxes, float NMS_THRESH)
 {
     std::sort(input_boxes.begin(), input_boxes.end(), [](BoxInfo a, BoxInfo b) { return a.score > b.score; });
     std::vector<float> vArea(input_boxes.size());
-    for (int i = 0; i < int(input_boxes.size()); ++i) 
+    for (int i = 0; i < int(input_boxes.size()); ++i)
     {
         vArea[i] = (input_boxes.at(i).x2 - input_boxes.at(i).x1 + 1)
             * (input_boxes.at(i).y2 - input_boxes.at(i).y1 + 1);
     }
-    for (int i = 0; i < int(input_boxes.size()); ++i) 
+    for (int i = 0; i < int(input_boxes.size()); ++i)
     {
-        for (int j = i + 1; j < int(input_boxes.size());) 
+        for (int j = i + 1; j < int(input_boxes.size());)
         {
             float xx1 = (std::max)(input_boxes[i].x1, input_boxes[j].x1);
             float yy1 = (std::max)(input_boxes[i].y1, input_boxes[j].y1);
@@ -178,15 +178,15 @@ void NanoDet::nms(std::vector<BoxInfo>& input_boxes, float NMS_THRESH)
             float h = (std::max)(float(0), yy2 - yy1 + 1);
             float inter = w * h;
             float ovr = inter / (vArea[i] + vArea[j] - inter);
-            if (ovr >= NMS_THRESH) 
+            if (ovr >= NMS_THRESH)
             {
                 input_boxes.erase(input_boxes.begin() + j);
                 vArea.erase(vArea.begin() + j);
             }
-            else 
+            else
             {
                 j++;
             }
         }
     }
-}
+}

demo_libtorch/nanodet_libtorch.h

@@ -2,7 +2,7 @@
 #include <opencv2/core/core.hpp>
 
 
-typedef struct BoxInfo 
+typedef struct BoxInfo
 {
     float x1;
     float y1;
@@ -30,4 +30,4 @@ class NanoDet
     int num_class_ = 80;
     int reg_max_ = 7;
 
-};
+};

demo_mnn/main.cpp

@@ -195,7 +195,7 @@ void draw_bboxes(const cv::Mat& bgr, const std::vector<BoxInfo>& bboxes, object_
         //fprintf(stderr, "%d = %.5f at %.2f %.2f %.2f %.2f\n", bbox.label, bbox.score,
         //    bbox.x1, bbox.y1, bbox.x2, bbox.y2);
 
-        cv::rectangle(image, cv::Rect(cv::Point((bbox.x1 - effect_roi.x) * width_ratio, (bbox.y1 - effect_roi.y) * height_ratio), 
+        cv::rectangle(image, cv::Rect(cv::Point((bbox.x1 - effect_roi.x) * width_ratio, (bbox.y1 - effect_roi.y) * height_ratio),
                                       cv::Point((bbox.x2 - effect_roi.x) * width_ratio, (bbox.y2 - effect_roi.y) * height_ratio)), color);
 
         char text[256];
@@ -250,7 +250,7 @@ int image_demo(NanoDet &detector, const char* imagepath)
         resize_uniform(image, resized_img, cv::Size(320, 320), effect_roi);
         std::vector<BoxInfo> results;
         detector.detect(resized_img, results);
-        
+
         #ifdef __SAVE_RESULT__
             std::string save_path = img_name.operator std::string();
             draw_bboxes(image, results, effect_roi, save_path.replace(3, 4, "results"));
@@ -319,7 +319,7 @@ int benchmark(NanoDet& detector)
 
         std::chrono::duration<double> elapsed = end - start;
         double time = elapsed.count();
-        if (i >= warm_up) 
+        if (i >= warm_up)
         {
             time_min = (std::min)(time_min, time);
             time_max = (std::max)(time_max, time);
@@ -369,4 +369,3 @@ int main(int argc, char** argv)
         }
     }
 }
-

demo_ncnn/CMakeLists.txt

@@ -30,7 +30,7 @@ include_directories(
 add_executable(nanodet_demo main.cpp nanodet.cpp)
 
 target_link_libraries(
-    nanodet_demo 
+    nanodet_demo
     ncnn
-    ${OpenCV_LIBS} 
-)
+    ${OpenCV_LIBS}
+)

demo_ncnn/README.md

@@ -49,7 +49,7 @@ Download Vulkan SDK from https://vulkan.lunarg.com/sdk/home
 Clone NCNN repository
 
 ``` shell script
-git clone --recursive https://github.com/Tencent/ncnn.git 
+git clone --recursive https://github.com/Tencent/ncnn.git
 ```
 
 Build NCNN following this tutorial: [Build for Linux / NVIDIA Jetson / Raspberry Pi](https://github.com/Tencent/ncnn/wiki/how-to-build#build-for-linux)