Added the generation of camera poses when the

timestamps are generated

event_camera_simulator/.gitignore

@@ -2,3 +2,4 @@
 esim_ros/scripts/exp_*
 *.swp
 *.autosave
+.clang-format

event_camera_simulator/esim/include/esim/esim/camera_simulator.hpp

@@ -11,12 +11,16 @@ namespace event_camera_simulator {
         ZE_POINTER_TYPEDEFS(ImageBuffer);
 
         struct ImageData {
-            ImageData(Image img, Time stamp, Duration exposure_time)
+            ImageData(
+                Image img, ImageRGB img_rgb, Time stamp, Duration exposure_time
+            )
                 : image(img),
+                  image_rgb(img_rgb),
                   stamp(stamp),
                   exposure_time(exposure_time) {}
 
             Image image;
+            ImageRGB image_rgb;
             Time stamp;
             Duration exposure_time; // timestamp since last image (0 if this is
                                     // the first image)
@@ -27,7 +31,7 @@ namespace event_camera_simulator {
         // Rolling image buffer of mazimum size 'buffer_size_ns'.
         ImageBuffer(Duration buffer_size_ns): buffer_size_ns_(buffer_size_ns) {}
 
-        void addImage(Time t, const Image& img);
+        void addImage(Time t, const Image& img, const ImageRGB& img_rgb);
 
         std::deque<ImageData> getRawBuffer() const {
             return data_;
@@ -63,7 +67,10 @@ namespace event_camera_simulator {
         }
 
         bool imageCallback(
-            const Image& img, Time time, const ImagePtr& camera_image
+            const Image& img,
+            const ImageRGB& img_rgb,
+            Time time,
+            const ImagePtr& camera_image
         );
 
       private:

event_camera_simulator/esim/src/camera_simulator.cpp

@@ -4,6 +4,7 @@
 #include <esim/esim/camera_simulator.hpp>
 #include <fstream>
 #include <iomanip>
+#include <iostream>
 #include <ostream>
 #include <ze/common/file_utils.hpp>
 
@@ -12,7 +13,8 @@ static std::ofstream exposures_file_;
 
 namespace event_camera_simulator {
 
-    void ImageBuffer::addImage(Time t, const Image& img) {
+    void
+    ImageBuffer::addImage(Time t, const Image& img, const ImageRGB& img_rgb) {
         if (!data_.empty()) {
             // Check that the image timestamps are monotonically increasing
             CHECK_GT(t, data_.back().stamp);
@@ -21,7 +23,9 @@ namespace event_camera_simulator {
         Duration exposure_time = data_.empty() ? 0 : t - data_.back().stamp;
         VLOG(2) << "Adding image to buffer with stamp: " << t
                 << " and exposure time " << exposure_time;
-        data_.push_back(ImageData(img.clone(), t, exposure_time));
+        data_.push_back(
+            ImageData(img.clone(), img_rgb.clone(), t, exposure_time)
+        );
 
         // Remove all the images with timestamp older than t - buffer_size_ns_
         auto first_valid_element = std::lower_bound(
@@ -40,12 +44,16 @@ namespace event_camera_simulator {
     }
 
     bool CameraSimulator::imageCallback(
-        const Image& img, Time time, const ImagePtr& camera_image
+        const Image& img,
+        const ImageRGB& img_rgb,
+        Time time,
+        const ImagePtr& camera_image
     ) {
         CHECK(camera_image);
         CHECK_EQ(camera_image->size(), img.size());
+        CHECK_EQ(img_rgb.size(), img.size());
 
-        buffer_->addImage(time, img);
+        buffer_->addImage(time, img, img_rgb);
 
         static const Time initial_time = time;
         if (time - initial_time < exposure_time_) {
@@ -55,10 +63,15 @@ namespace event_camera_simulator {
             return false;
         }
 
+        ImageRGB rgb(img.rows, img.cols, CV_32FC3);
+        rgb.setTo(0.);
+
         // average all the images in the buffer to simulate motion blur
         camera_image->setTo(0);
         ze::real_t denom = 0.;
         for (const ImageBuffer::ImageData& img : buffer_->getRawBuffer()) {
+            rgb +=
+                ze::nanosecToMillisecTrunc(img.exposure_time) * img.image_rgb;
             *camera_image +=
                 ze::nanosecToMillisecTrunc(img.exposure_time) * img.image;
             denom += ze::nanosecToMillisecTrunc(img.exposure_time);
@@ -70,7 +83,7 @@ namespace event_camera_simulator {
         ss << hdr_output_folder << "/frames/frame_" << std::setfill('0')
            << std::setw(5) << frame_number++ << ".exr";
         std::string frame_path = ss.str();
-        cv::imwrite(frame_path, *camera_image);
+        cv::imwrite(frame_path, rgb);
 
         // ze::openOutputFileStream(
         //     ze::joinPath(output_folder, "exposures.csv"),

event_camera_simulator/esim/src/simulator.cpp

@@ -46,6 +46,7 @@ namespace event_camera_simulator {
                     camera_simulator_success =
                         camera_simulators_[i].imageCallback(
                             *sim_data.images[i],
+                            *sim_data.images_rgb[i],
                             time,
                             corrupted_camera_images_[i]
                         );

event_camera_simulator/esim_common/include/esim/common/types.hpp

@@ -42,12 +42,15 @@ namespace event_camera_simulator {
     using Duration = ze::uint64_t;
     using Image = cv::Mat_<ImageFloatType>;
     using ImagePtr = std::shared_ptr<Image>;
+    using ImageRGB = cv::Mat;
+    using ImageRGBPtr = std::shared_ptr<ImageRGB>;
     using Depthmap = cv::Mat_<ImageFloatType>;
     using OpticFlow = cv::Mat_<cv::Vec<ImageFloatType, 2>>;
     using OpticFlowPtr = std::shared_ptr<OpticFlow>;
     using DepthmapPtr = std::shared_ptr<Depthmap>;
 
     using ImagePtrVector = std::vector<ImagePtr>;
+    using ImageRGBPtrVector = std::vector<ImageRGBPtr>;
     using DepthmapPtrVector = std::vector<DepthmapPtr>;
     using OpticFlowPtrVector = std::vector<OpticFlowPtr>;
 
@@ -98,6 +101,7 @@ namespace event_camera_simulator {
 
         //! Camera images.
         ImagePtrVector images;
+        ImageRGBPtrVector images_rgb;
 
         //! Depth maps.
         DepthmapPtrVector depthmaps;

event_camera_simulator/esim_data_provider/src/data_provider_from_folder.cpp

@@ -1,4 +1,11 @@
+#include "esim/imp_multi_objects_2d/object.hpp"
+#include "opencv2/imgcodecs.hpp"
+#include "opencv2/imgproc.hpp"
+#include "opencv2/imgproc/types_c.h"
+
+#include <cstdint>
 #include <esim/data_provider/data_provider_from_folder.hpp>
+#include <iostream>
 #include <opencv2/imgcodecs/imgcodecs.hpp>
 #include <ze/cameras/camera_impl.hpp>
 #include <ze/common/file_utils.hpp>
@@ -40,6 +47,11 @@ namespace event_camera_simulator {
         sim_data_.images.emplace_back(ImagePtr(new Image(
             cv::Size(camera_rig_->at(0).width(), camera_rig_->at(0).height())
         )));
+        sim_data_.images_rgb.emplace_back(ImageRGBPtr(new ImageRGB(
+            camera_rig_->at(0).width(),
+            camera_rig_->at(0).height(),
+            CV_32FC3
+        )));
 
         sim_data_.camera_rig = camera_rig_;
         sim_data_.images_updated = true;
@@ -74,12 +86,17 @@ namespace event_camera_simulator {
 
             const std::string& path_to_image =
                 ze::joinPath(path_to_data_folder_, items[1]);
-            cv::Mat image = cv::imread(path_to_image, 0);
+            cv::Mat image = cv::imread(path_to_image);
             CHECK(image.data)
                 << "Could not load image from file: " << path_to_image;
 
             VLOG(3) << "Read image from file: " << path_to_image;
-            image.convertTo(
+            image.copyTo(*sim_data_.images_rgb[0]);
+
+            cv::Mat image_gray = image;
+            cv::cvtColor(image, image_gray, CV_BGR2GRAY);
+
+            image_gray.convertTo(
                 *sim_data_.images[0],
                 cv::DataType<ImageFloatType>::type,
                 1. / 255.

event_camera_simulator/esim_data_provider/src/data_provider_online_render.cpp

@@ -103,11 +103,15 @@ namespace event_camera_simulator {
             );
 
             sim_data_.images.emplace_back(ImagePtr(new Image(size)));
+            sim_data_.images_rgb.emplace_back(
+                ImageRGBPtr(new ImageRGB(size, CV_32FC3))
+            );
             sim_data_.depthmaps.emplace_back(DepthmapPtr(new Depthmap(size)));
             sim_data_.optic_flows.emplace_back(OpticFlowPtr(new OpticFlow(size))
             );
 
             sim_data_.images[i]->setTo(0);
+            sim_data_.images_rgb[i]->setTo(0);
             sim_data_.depthmaps[i]->setTo(0);
             sim_data_.optic_flows[i]->setTo(0);
         }
@@ -241,6 +245,7 @@ namespace event_camera_simulator {
                         sim_data_.groundtruth.linear_velocity_obj_,
                         sim_data_.groundtruth.angular_velocity_obj_,
                         sim_data_.images[i],
+                        sim_data_.images_rgb[i],
                         sim_data_.depthmaps[i],
                         sim_data_.optic_flows[i]
                     );
@@ -249,6 +254,7 @@ namespace event_camera_simulator {
                         sim_data_.groundtruth.T_W_B * camera_rig_->T_B_C(i),
                         sim_data_.groundtruth.T_W_OBJ_,
                         sim_data_.images[i],
+                        sim_data_.images_rgb[i],
                         sim_data_.depthmaps[i]
                     );
                 }

event_camera_simulator/esim_rendering/include/esim/rendering/renderer_base.hpp

@@ -19,6 +19,7 @@ namespace event_camera_simulator {
             const Transformation& T_W_C,
             const std::vector<Transformation>& T_W_OBJ,
             const ImagePtr& out_image,
+            const ImageRGBPtr& out_image_rgb,
             const DepthmapPtr& out_depthmap
         ) const = 0;
 
@@ -36,6 +37,7 @@ namespace event_camera_simulator {
             const std::vector<LinearVelocity>& linear_velocity_obj,
             const std::vector<AngularVelocity>& angular_velocity_obj,
             const ImagePtr& out_image,
+            const ImageRGBPtr& out_image_rgb,
             const DepthmapPtr& out_depthmap,
             const OpticFlowPtr& optic_flow_map
         ) const {}

event_camera_simulator/esim_ros/cfg/from_folder.conf

@@ -1,7 +1,7 @@
  --data_source=2
 
- --path_to_output_bag=/tmp/out.bag
+ # --path_to_output_bag=/tmp/out.bag
- --path_to_data_folder=/tmp/tests/frames
+ --path_to_data_folder=/home/arno/Videos/sponze/frames
  --hdr_output_folder=/home/arno/sim_ws/out
 
  --ros_publisher_frame_rate=60

event_camera_simulator/esim_ros/scripts/generate_stamps_file.py

@@ -1,25 +1,45 @@
 import argparse
 from os import listdir
 from os.path import join
+import numpy as np
+from scipy import interpolate
+import csv
 
 
-if __name__ == "__main__":
+if __name__ == '__main__':
 
     parser = argparse.ArgumentParser(
-        description='Generate "images.csv" for ESIM DataProviderFromFolder')
+        description='Generate "images.csv" for ESIM DataProviderFromFolder'
+    )
 
-    parser.add_argument('-i', '--input_folder', default=None, type=str,
+    parser.add_argument(
-                        help="folder containing the images")
+        '-i',
-    parser.add_argument('-r', '--framerate', default=1000, type=float,
+        '--input_folder',
-                        help="video framerate, in Hz")
+        default=None,
+        type=str,
+        help='folder containing the images',
+    )
+    parser.add_argument(
+        '-r',
+        '--framerate',
+        default=1000,
+        type=float,
+        help='video framerate, in Hz',
+    )
 
     args = parser.parse_args()
 
     images = sorted(
-        [f for f in listdir(args.input_folder) if f.endswith('.png')])
+        [
+            f
+            for f in listdir(args.input_folder)
+            if f.endswith('.exr') or f.endswith('.png')
+        ]
+    )
 
-    print('Will write file: {} with framerate: {} Hz'.format(
+    print(
-        join(args.input_folder, 'images.csv'), args.framerate))
+        f"Will write timestamps in file: {join(args.input_folder, 'images.csv')} with framerate: {args.framerate} Hz"
+    )
     stamp_nanoseconds = 1
     dt_nanoseconds = int((1.0 / args.framerate) * 1e9)
     with open(join(args.input_folder, 'images.csv'), 'w') as f:
@@ -28,3 +48,38 @@ if __name__ == "__main__":
             stamp_nanoseconds += dt_nanoseconds
 
     print('Done!')
+    print(
+        f"Will write camera poses in file: {join(args.input_folder, 'images.csv')}"
+    )
+    fields = ['frame number', 'x', 'y', 'z', 'roll', 'pitch', 'yaw']
+    with open(join(args.input_folder, 'poses.csv'), 'w') as poses:
+        writer = csv.DictWriter(poses, fieldnames=fields)
+        writer.writeheader()
+
+        with open(join(args.input_folder, 'keyframes.csv'), 'r') as csv_file:
+            reader = csv.DictReader(csv_file)
+            for n, row in enumerate(reader):
+                for field in row:
+                    row[field] = float(row[field])
+                if n == 0:
+                    prev = row
+                    continue
+                start, end = prev['frame number'], row['frame number']
+                time = np.arange(start, end)
+
+                data = []
+                for t in time:
+                    data.append({f: 0 for f in fields})
+                    data[-1]['frame number'] = int(t)
+
+                for var in fields[1:]:
+                    f = interpolate.CubicHermiteSpline(
+                        [start, end], [prev[var], row[var]], [0, 0]
+                    )
+
+                    # verify(f, var, start, end)
+                    for i, value in enumerate(f(time)):
+                        data[i][var] = value
+                    prev = row
+                writer.writerows(data)
+        print('Done!')

event_camera_simulator/esim_visualization/src/hdr_publisher.cpp

@@ -100,14 +100,14 @@ namespace event_camera_simulator {
     }
 
     void HdrPublisher::imageCallback(const ImagePtrVector& images, Time t) {
-        CHECK_EQ(images.size(), 1);
+        // CHECK_EQ(images.size(), 1);
-        static uint frame_number = 0;
+        // static uint frame_number = 0;
-        std::stringstream ss;
+        // std::stringstream ss;
-        ss << output_folder_ << "/frames/frame_" << std::setfill('0')
+        // ss << output_folder_ << "/frames/frame_" << std::setfill('0')
-           << std::setw(5) << frame_number++ << ".exr";
+        //    << std::setw(5) << frame_number++ << ".exr";
-        std::string frame_path = ss.str();
+        // std::string frame_path = ss.str();
-
+        //
-        cv::imwrite(frame_path, *images[0]);
+        // cv::imwrite(frame_path, *images[0]);
     }
 
     void HdrPublisher::eventsCallback(const EventsVector& events) {

event_camera_simulator/imp/imp_opengl_renderer/include/esim/imp_opengl_renderer/opengl_renderer.hpp

@@ -24,8 +24,9 @@ namespace event_camera_simulator {
             const Transformation& T_W_C,
             const std::vector<Transformation>& T_W_OBJ,
             const ImagePtr& out_image,
+            const ImageRGBPtr& out_image_rgb,
             const DepthmapPtr& out_depthmap
-        ) const;
+        ) const override;
 
         //! Returns true if the rendering engine can compute optic flow, false
         //! otherwise
@@ -43,6 +44,7 @@ namespace event_camera_simulator {
             const std::vector<LinearVelocity>& linear_velocity_obj,
             const std::vector<AngularVelocity>& angular_velocity_obj,
             const ImagePtr& out_image,
+            const ImageRGBPtr& out_image_rgb,
             const DepthmapPtr& out_depthmap,
             const OpticFlowPtr& optic_flow_map
         ) const override;

event_camera_simulator/imp/imp_opengl_renderer/src/opengl_renderer.cpp

@@ -17,6 +17,7 @@
 #include <esim/imp_opengl_renderer/opengl_renderer.hpp>
 
 #include <glad/glad.h>
+#include <iomanip>
 #include <learnopengl/shader.h>
 #include <learnopengl/model.h>
 #include <GLFW/glfw3.h>
@@ -314,18 +315,23 @@ namespace event_camera_simulator {
         const Transformation& T_W_C,
         const std::vector<Transformation>& T_W_OBJ,
         const ImagePtr& out_image,
+        const ImageRGBPtr& out_image_rgb,
         const DepthmapPtr& out_depthmap
     ) const {
         CHECK(is_initialized_) << "Called render() but the renderer was not "
                                   "initialized yet. Have you "
                                   "first called setCamera()?";
         CHECK(out_image);
+        CHECK(out_image_rgb);
         CHECK(out_depthmap);
         CHECK_EQ(out_image->cols, width_);
+        CHECK_EQ(out_image_rgb->cols, width_);
         CHECK_EQ(out_image->rows, height_);
+        CHECK_EQ(out_image_rgb->rows, height_);
         CHECK_EQ(out_depthmap->cols, width_);
         CHECK_EQ(out_depthmap->rows, height_);
         CHECK_EQ(out_image->type(), CV_32F);
+        CHECK_EQ(out_image_rgb->type(), CV_32FC3);
         CHECK_EQ(out_depthmap->type(), CV_32F);
 
         // draw to our framebuffer instead of screen
@@ -413,6 +419,29 @@ namespace event_camera_simulator {
 
         // read out what we just rendered
         cv::Mat img_color(height_, width_, CV_8UC3);
+        {
+            float pixels[height_ * width_ * 3];
+            glReadPixels(
+                0,
+                0,
+                img_color.cols,
+                img_color.rows,
+                GL_RGB,
+                GL_FLOAT,
+                pixels
+            );
+
+            cv::Mat rgb(height_, width_, CV_32FC3, pixels);
+            rgb.copyTo(*out_image_rgb);
+        }
+        // static uint frame_number = 0;
+
+        // std::stringstream ss;
+        // ss << "/tmp/tests"
+        //    << "/frames/frame_" << std::setfill('0') << std::setw(5)
+        //    << frame_number++ << ".exr";
+        // cv::imwrite(ss.str(), rgb);
+
         glPixelStorei(GL_PACK_ALIGNMENT, (img_color.step & 3) ? 1 : 4);
         glPixelStorei(
             GL_PACK_ROW_LENGTH,
@@ -480,10 +509,11 @@ namespace event_camera_simulator {
         const std::vector<LinearVelocity>& linear_velocity_obj,
         const std::vector<AngularVelocity>& angular_velocity_obj,
         const ImagePtr& out_image,
+        const ImageRGBPtr& out_image_rgb,
         const DepthmapPtr& out_depthmap,
         const OpticFlowPtr& optic_flow_map
     ) const {
-        render(T_W_C, T_W_OBJ, out_image, out_depthmap);
+        render(T_W_C, T_W_OBJ, out_image, out_image_rgb, out_depthmap);
 
         // draw to our optic flow framebuffer instead of screen
         glBindFramebuffer(GL_FRAMEBUFFER, fbo_of);

event_camera_simulator/imp/imp_panorama_renderer/include/esim/imp_panorama_renderer/panorama_renderer.hpp

@@ -26,6 +26,7 @@ namespace event_camera_simulator {
             const Transformation& T_W_C,
             const std::vector<Transformation>& T_W_OBJ,
             const ImagePtr& out_image,
+            const ImageRGBPtr& out_image_rgb,
             const DepthmapPtr& out_depthmap
         ) const override {
             render(T_W_C, out_image, out_depthmap);

event_camera_simulator/imp/imp_planar_renderer/include/esim/imp_planar_renderer/planar_renderer.hpp

@@ -31,6 +31,7 @@ namespace event_camera_simulator {
             const Transformation& T_W_C,
             const std::vector<Transformation>& T_W_OBJ,
             const ImagePtr& out_image,
+            const ImageRGBPtr& out_image_rbg,
             const DepthmapPtr& out_depthmap
         ) const override {
             render(T_W_C, out_image, out_depthmap);

event_camera_simulator/imp/imp_unrealcv_renderer/include/esim/imp_unrealcv_renderer/unrealcv_renderer.hpp

@@ -21,8 +21,9 @@ namespace event_camera_simulator {
             const Transformation& T_W_C,
             const std::vector<Transformation>& T_W_OBJ,
             const ImagePtr& out_image,
+            const ImageRGBPtr& out_image_rgb,
             const DepthmapPtr& out_depthmap
-        ) const {
+        ) const override {
             render(T_W_C, out_image, out_depthmap);
         }