Perception: Fix inference tools BUILD.

modules/perception/inference/tools/BUILD

+load("//tools:cpplint.bzl", "cpplint")
+
+package(default_visibility = ["//visibility:public"])
+
+cc_binary(
+    name = "cal_table_generator",
+    srcs = ["cal_table_generator.cc"],
+    deps = [
+        "//modules/perception/inference:inference_factory_lib",
+        "//modules/perception/inference:inference_lib",
+        "//modules/perception/inference/tensorrt:batch_stream",
+        "//modules/perception/inference/tensorrt:entropy_calibrator",
+        "//modules/perception/inference/tensorrt:rt_net",
+        "//modules/perception/inference/utils:inference_util_lib",
+    ],
+)
+
+cc_binary(
+    name = "denseline_sample",
+    srcs = ["denseline_sample.cc"],
+    deps = [
+        "//modules/perception/inference:inference_factory_lib",
+        "//modules/perception/inference:inference_lib",
+        "//modules/perception/inference/tensorrt:batch_stream",
+        "//modules/perception/inference/tensorrt:entropy_calibrator",
+        "//modules/perception/inference/tensorrt:rt_net",
+        "//modules/perception/inference/utils:inference_util_lib",
+    ],
+)
+
+cc_binary(
+    name = "lane_sample",
+    srcs = ["lane_sample.cc"],
+    deps = [
+        "//modules/perception/inference:inference_factory_lib",
+        "//modules/perception/inference:inference_lib",
+        "//modules/perception/inference/tensorrt:batch_stream",
+        "//modules/perception/inference/tensorrt:entropy_calibrator",
+        "//modules/perception/inference/tensorrt:rt_net",
+        "//modules/perception/inference/utils:inference_util_lib",
+    ],
+)
+
+cc_binary(
+    name = "yolo_sample",
+    srcs = ["yolo_sample.cc"],
+    deps = [
+        "//modules/perception/inference:inference_factory_lib",
+        "//modules/perception/inference:inference_lib",
+        "//modules/perception/inference/tensorrt:batch_stream",
+        "//modules/perception/inference/tensorrt:entropy_calibrator",
+        "//modules/perception/inference/tensorrt:rt_net",
+        "//modules/perception/inference/utils:inference_util_lib",
+    ],
+)
+
+cpplint()

modules/perception/inference/tools/BUILD.txt

-# DO NOT MODIFY
-# THIS FILE IS GENERATED AUTOMATICALLY TO HELP WRITE BUILD FILE
-# DELETE THIS FILE WHEN YOU HAVE CREATED THE COMPLETE BUILD FILE
-load("//tools:cpplint.bzl", "cpplint")
-
-package(default_visibility = ["//visibility:public"])
-
-cc_library(
-    name = "cal_table_generator",
-    srcs = [
-        "cal_table_generator.cc",
-    ],
-    deps = [
-    ],
-)
-
-cc_library(
-    name = "denseline_sample",
-    srcs = [
-        "denseline_sample.cc",
-    ],
-    deps = [
-    ],
-)
-
-cc_library(
-    name = "lane_sample",
-    srcs = [
-        "lane_sample.cc",
-    ],
-    deps = [
-    ],
-)
-
-cc_library(
-    name = "yolo_sample",
-    srcs = [
-        "yolo_sample.cc",
-    ],
-    deps = [
-    ],
-)
-
-cpplint()

modules/perception/inference/tools/cal_table_generator.cc

@@ -13,10 +13,12 @@
  * See the License for the specific language governing permissions and
  * limitations under the License.
  *****************************************************************************/
-#include <gflags/gflags.h>
 #include <iostream>
+
+#include "gflags/gflags.h"
 #include "opencv2/opencv.hpp"
 
+#include "cybertron/common/log.h"
 #include "modules/perception/inference/inference.h"
 #include "modules/perception/inference/tensorrt/batch_stream.h"
 #include "modules/perception/inference/tensorrt/entropy_calibrator.h"
@@ -46,9 +48,9 @@ DEFINE_bool(hwc_input, true, "input blob is hwc order.");
 
 int evaluate_image_list() {
   CHECK_EQ(FLAGS_image_channel_num, 3);
-  int height = FLAGS_height;
-  int width = FLAGS_width;
-  int count = FLAGS_image_channel_num * width * height;
+  const int height = FLAGS_height;
+  const int width = FLAGS_width;
+  const int count = FLAGS_image_channel_num * width * height;
 
   std::ifstream fin;
   fin.open(FLAGS_test_list, std::ifstream::in);
@@ -69,8 +71,7 @@ int evaluate_image_list() {
   }
   fin.close();
 
-  std::string out_file;
-  out_file = FLAGS_batch_root + "/Batch0";
+  std::string out_file = FLAGS_batch_root + "/Batch0";
   std::ofstream out_car(out_file, std::ofstream::out | std::ofstream::binary);
   // std::ofstream out_car(out_file, std::ofstream::out);
   if (!out_car.is_open()) {
@@ -79,7 +80,7 @@ int evaluate_image_list() {
   }
   std::vector<float> cpu_data(count);
   // main loop
-  for (int i = 0; i < img_list.size(); i++) {
+  for (size_t i = 0; i < img_list.size(); i++) {
     std::string image_path = img_list[i] + FLAGS_image_ext;
     cv::Mat img = cv::imread(image_path, CV_LOAD_IMAGE_COLOR);
     cv::Mat img_org;
@@ -157,8 +158,6 @@ int evaluate_image_list() {
 int main(int argc, char **argv) {
   google::ParseCommandLineFlags(&argc, &argv, true);
 
-  std::string proto_file = FLAGS_proto_file;
-  std::string weight_file = FLAGS_weight_file;
   if (FLAGS_gen_batch) {
     evaluate_image_list();
   }
@@ -173,7 +172,8 @@ int main(int argc, char **argv) {
       (new nvinfer1::Int8EntropyCalibrator(stream, 0, true,
                                            FLAGS_cal_table_root));
   apollo::perception::inference::RTNet *rt_net =
-      new apollo::perception::inference::RTNet(proto_file, weight_file, outputs,
+      new apollo::perception::inference::RTNet(FLAGS_proto_file,
+                                               FLAGS_weight_file, outputs,
                                                inputs, calibrator);
   rt_net->Init(std::map<std::string, std::vector<int>>());
   rt_net->Infer();

modules/perception/inference/tools/denseline_sample.cc

@@ -14,9 +14,13 @@
  * limitations under the License.
  *****************************************************************************/
 #include <iostream>
-#include "/usr/include/opencv2/opencv.hpp"
-#include "/usr/local/include/gflags/gflags.h"
+
+#include "gflags/gflags.h"
+#include "opencv2/opencv.hpp"
+
+#include "cybertron/common/log.h"
 #include "modules/perception/inference/inference.h"
+#include "modules/perception/inference/inference_factory.h"
 #include "modules/perception/inference/tensorrt/batch_stream.h"
 #include "modules/perception/inference/tensorrt/entropy_calibrator.h"
 #include "modules/perception/inference/tensorrt/rt_net.h"
@@ -43,27 +47,27 @@ int main(int argc, char **argv) {
   google::ParseCommandLineFlags(&argc, &argv, true);
   std::vector<float> output_data_vec1;
 
-  int height = FLAGS_height;
-  int width = FLAGS_width;
-  int count = FLAGS_image_channel * width * height;
+  const int height = FLAGS_height;
+  const int width = FLAGS_width;
+  const int count = FLAGS_image_channel * width * height;
 
-  std::string proto_file = "./denseline_parser/deploy.prototxt";
-  std::string weight_file = "./denseline_parser/deploy.caffemodel";
-  std::string model_root = "./denseline_parser/";
+  const std::string proto_file = "./denseline_parser/deploy.prototxt";
+  const std::string weight_file = "./denseline_parser/deploy.caffemodel";
+  const std::string model_root = "./denseline_parser/";
 
   cudaDeviceProp prop;
   cudaGetDeviceProperties(&prop, 0);
-  LOG(INFO) << prop.name;
+  AINFO << prop.name;
 
   apollo::perception::inference::Inference *rt_net;
-  std::string input_blob_name = "data";
+  const std::string input_blob_name = "data";
   std::vector<std::string> inputs{input_blob_name};
 
   std::vector<std::string> outputs;
   apollo::perception::inference::load_data<std::string>(FLAGS_names_file,
                                                         &outputs);
   for (auto name : outputs) {
-    LOG(INFO) << "outputs name: " << name;
+    AINFO << "outputs name: " << name;
   }
 
   // std::shared_ptr<nvinfer1::Int8EntropyCalibrator> calibrator = nullptr;
@@ -74,11 +78,11 @@ int main(int argc, char **argv) {
     // apollo::perception::inference::BatchStream stream(2, 50, "./batches/");
     // calibrator
     //    .reset(new nvinfer1::Int8EntropyCalibrator(stream, 0, true, "./"));
-    LOG(INFO) << "int8";
+    AINFO << "int8";
     rt_net = new apollo::perception::inference::RTNet(
         proto_file, weight_file, outputs, inputs, calibrator);
   } else {
-    LOG(INFO) << "fp32";
+    AINFO << "fp32";
     rt_net = apollo::perception::inference::CreateInferenceByName(
         "RTNet", proto_file, weight_file, outputs, inputs);
   }
@@ -95,7 +99,7 @@ int main(int argc, char **argv) {
 
   for (auto image_file : image_lists) {
     cv::Mat img = cv::imread(FLAGS_image_root + image_file + FLAGS_image_ext);
-    LOG(INFO) << img.channels();
+    AINFO << img.channels();
     cv::Rect roi(0, FLAGS_offset_y, img.cols, img.rows - FLAGS_offset_y);
     cv::Mat img_roi = img(roi);
     img_roi.copyTo(img);
@@ -124,26 +128,23 @@ int main(int argc, char **argv) {
       std::vector<float> tmp_vec(blob->cpu_data(),
                                  blob->cpu_data() + blob->count());
       // if(output_name=="conv2" || output_name == "conv1") {
-      if (1) {
-        int cnt = 0;
-        LOG(INFO) << output_name << " " << blob->channels() << " "
-                  << blob->height() << " " << blob->width();
-        double sum = 0;
-        for (int i = 0; i < blob->count(); ++i) {
-          sum += blob->cpu_data()[i];
-          if (i < 100) {
-            LOG(INFO) << blob->cpu_data()[i];
-          }
+      AINFO << output_name << " " << blob->channels() << " "
+            << blob->height() << " " << blob->width();
+      double sum = 0;
+      for (int i = 0; i < blob->count(); ++i) {
+        sum += blob->cpu_data()[i];
+        if (i < 100) {
+          AINFO << blob->cpu_data()[i];
         }
-        LOG(INFO) << output_name << ", sum : " << std::endl;
       }
+      AINFO << output_name << ", sum : " << std::endl;
+      // end of if
+
       output_data_vec.insert(output_data_vec.end(), tmp_vec.begin(),
                              tmp_vec.end());
     }
   }
   apollo::perception::inference::write_result(FLAGS_res_dir, output_data_vec);
-
-  LOG(INFO) << "Hello, World!";
   delete rt_net;
   // rt_net has deletle calibrator?
   // delete calibrator;

modules/perception/inference/tools/lane_sample.cc

@@ -14,9 +14,13 @@
  * limitations under the License.
  *****************************************************************************/
 #include <iostream>
-#include "/usr/include/opencv2/opencv.hpp"
-#include "/usr/local/include/gflags/gflags.h"
+
+#include "gflags/gflags.h"
+#include "opencv2/opencv.hpp"
+
+#include "cybertron/common/log.h"
 #include "modules/perception/inference/inference.h"
+#include "modules/perception/inference/inference_factory.h"
 #include "modules/perception/inference/tensorrt/batch_stream.h"
 #include "modules/perception/inference/tensorrt/entropy_calibrator.h"
 #include "modules/perception/inference/tensorrt/rt_net.h"
@@ -29,6 +33,7 @@ DEFINE_string(test_list, "image_list.txt", "path of image list");
 DEFINE_string(image_root, "./images/", "path of image dir");
 DEFINE_string(image_ext, ".jpg", "path of image ext");
 DEFINE_string(res_dir, "./result.dat", "path of result");
+
 int main(int argc, char **argv) {
   google::ParseCommandLineFlags(&argc, &argv, true);
   std::vector<float> output_data_vec1;
@@ -39,16 +44,16 @@ int main(int argc, char **argv) {
   color_table.push_back(cv::Scalar(255, 0, 0));     // for mask value = 3
   color_table.push_back(cv::Scalar(240, 32, 160));  // for mask value = 4
 
-  std::string proto_file = "./lane_parser/caffe.pt";
-  std::string weight_file = "./lane_parser/caffe.caffemodel";
-  std::string model_root = "./lane_parser/";
+  const std::string proto_file = "./lane_parser/caffe.pt";
+  const std::string weight_file = "./lane_parser/caffe.caffemodel";
+  const std::string model_root = "./lane_parser/";
 
   cudaDeviceProp prop;
   cudaGetDeviceProperties(&prop, 0);
-  LOG(INFO) << prop.name;
+  AINFO << prop.name;
 
   apollo::perception::inference::Inference *rt_net;
-  std::string input_blob_name = "data";
+  const std::string input_blob_name = "data";
   std::vector<std::string> inputs{"data"};
 
   std::vector<std::string> outputs{"loc_pred", "obj_pred", "cls_pred",
@@ -72,9 +77,9 @@ int main(int argc, char **argv) {
     rt_net = apollo::perception::inference::CreateInferenceByName(
         "RTNet", proto_file, weight_file, outputs, inputs);
   }
-  int height = 608;
-  int width = 1024;
-  int offset_y = 0;
+  const int height = 608;
+  const int width = 1024;
+  const int offset_y = 0;
   std::vector<int> shape = {1, 3, height, width};
   std::map<std::string, std::vector<int>> shape_map{{input_blob_name, shape}};
 
@@ -95,11 +100,10 @@ int main(int argc, char **argv) {
     img_roi.copyTo(img);
     cv::resize(img, img, cv::Size(width, height));
 
-    int count = 1 * width * height;
+    const int count = 1 * width * height;
     float *input = new float[count];
     for (int i = 0; i < count; i++) {
       input[i] = (img.data[i] - 128) * 0.0078125;
-      ;
     }
     cudaMemcpy(input_blob->mutable_gpu_data(), input, count * sizeof(float),
                cudaMemcpyHostToDevice);
@@ -111,29 +115,7 @@ int main(int argc, char **argv) {
       auto blob = rt_net->get_blob(output_name);
       std::vector<float> tmp_vec(blob->cpu_data(),
                                  blob->cpu_data() + blob->count());
-#if 0
-      if (output_name == "output") {
-        int cnt = 0;
-        std::cout << output_name << " " << blob->channels() << " "
-                  << blob->height() << " " << blob->width() << std::endl;
-        for (int c = 0; c < blob->channels(); c++) {
-          for (int j = 0; j < blob->height(); j++) {
-            for (int w = 0; w < blob->width(); w++) {
-              int offset = blob->offset(0, c, j, w);
-              cnt++;
-              std::cout << blob->cpu_data()[offset] << std::endl;
-            }
-          }
-        }
-        std::cout << std::endl;
-      }
-
-#endif
       if (output_name == "output") {
-        int output_height = blob->height();
-        int output_width = blob->width();
-        int output_channels = blob->channels();
-        int spatial_dim = output_height * output_width;
         const float *output_data = blob->cpu_data();
         for (int h = 0; h < img.rows; h++) {
           for (int w = 0; w < img.cols; w++) {
@@ -152,9 +134,6 @@ int main(int argc, char **argv) {
   }
 
   apollo::perception::inference::write_result(FLAGS_res_dir, output_data_vec);
-
-  std::cout << std::endl;
-  std::cout << "Hello, World!" << std::endl;
   delete rt_net;
 
   return 0;

modules/perception/inference/tools/yolo_sample.cc

@@ -14,9 +14,12 @@
  * limitations under the License.
  *****************************************************************************/
 #include <iostream>
-#include "/usr/include/opencv2/opencv.hpp"
-#include "/usr/local/include/gflags/gflags.h"
+
+#include "gflags/gflags.h"
+#include "opencv2/opencv.hpp"
+
 #include "modules/perception/inference/inference.h"
+#include "modules/perception/inference/inference_factory.h"
 #include "modules/perception/inference/tensorrt/batch_stream.h"
 #include "modules/perception/inference/tensorrt/entropy_calibrator.h"
 #include "modules/perception/inference/tensorrt/rt_net.h"
@@ -29,12 +32,13 @@ DEFINE_string(test_list, "image_list.txt", "path of image list");
 DEFINE_string(image_root, "./images/", "path of image dir");
 DEFINE_string(image_ext, ".jpg", "path of image ext");
 DEFINE_string(res_dir, "./result.dat", "path of result");
+
 int main(int argc, char **argv) {
   google::ParseCommandLineFlags(&argc, &argv, true);
 
-  std::string proto_file = "./yolo/caffe.pt";
-  std::string weight_file = "./yolo/caffe.caffemodel";
-  std::string model_root = "./yolo";
+  const std::string proto_file = "./yolo/caffe.pt";
+  const std::string weight_file = "./yolo/caffe.caffemodel";
+  const std::string model_root = "./yolo";
 
   apollo::perception::inference::Inference *rt_net;
   std::vector<std::string> outputs{"loc_pred", "obj_pred", "cls_pred",
@@ -55,9 +59,9 @@ int main(int argc, char **argv) {
     rt_net = apollo::perception::inference::CreateInferenceByName(
         "RTNet", proto_file, weight_file, outputs, inputs);
   }
-  int height = 576;
-  int width = 1440;
-  int offset_y = 312;
+  const int height = 576;
+  const int width = 1440;
+  const int offset_y = 312;
   std::vector<int> shape = {1, height, width, 3};
   std::map<std::string, std::vector<int>> shape_map{{"data", shape}};
   rt_net->Init(shape_map);
@@ -74,7 +78,7 @@ int main(int argc, char **argv) {
     img_roi.copyTo(img);
     cv::resize(img, img, cv::Size(width, height));
     auto input = rt_net->get_blob("data")->mutable_cpu_data();
-    int count = 3 * width * height;
+    const int count = 3 * width * height;
     for (int i = 0; i < count; i++) {
       input[i] = img.data[i];
     }
@@ -84,23 +88,13 @@ int main(int argc, char **argv) {
     for (auto output_name : outputs) {
       std::vector<int> shape;
       auto blob = rt_net->get_blob(output_name);
-      int size = blob->count();
+      const int size = blob->count();
       std::vector<float> tmp_vec(blob->cpu_data(), blob->cpu_data() + size);
-#if 0
-      std::cout << output_name << " " << std::endl;
-      for (int i = 0; i < 500; i++) {
-        std::cout << tmp_vec[i] << " ";
-      }
-      std::cout << std::endl;
-#endif
       output_data_vec.insert(output_data_vec.end(), tmp_vec.begin(),
                              tmp_vec.end());
     }
   }
   apollo::perception::inference::write_result(FLAGS_res_dir, output_data_vec);
-
-  std::cout << std::endl;
-  std::cout << "Hello, World!" << std::endl;
   delete rt_net;
   return 0;
 }