dlib/examples/dnn_instance_segmentation_e...

191 lines
7.3 KiB
C++
Raw Normal View History

Instance segmentation (#1918) * Add instance segmentation example - first version of training code * Add MMOD options; get rid of the cache approach, and instead load all MMOD rects upfront * Improve console output * Set filter count * Minor tweaking * Inference - first version, at least compiles! * Ignore overlapped boxes * Ignore even small instances * Set overlaps_ignore * Add TODO remarks * Revert "Set overlaps_ignore" This reverts commit 65adeff1f89af62b10c691e7aa86c04fc358d03e. * Set result size * Set label image size * Take ignore-color into account * Fix the cropping rect's aspect ratio; also slightly expand the rect * Draw the largest findings last * Improve masking of the current instance * Add some perturbation to the inputs * Simplify ground-truth reading; fix random cropping * Read even class labels * Tweak default minibatch size * Learn only one class * Really train only instances of the selected class * Remove outdated TODO remark * Automatically skip images with no detections * Print to console what was found * Fix class index problem * Fix indentation * Allow to choose multiple classes * Draw rect in the color of the corresponding class * Write detector window classes to ostream; also group detection windows by class (when ostreaming) * Train a separate instance segmentation network for each classlabel * Use separate synchronization file for each seg net of each class * Allow more overlap * Fix sorting criterion * Fix interpolating the predicted mask * Improve bilinear interpolation: if output type is an integer, round instead of truncating * Add helpful comments * Ignore large aspect ratios; refactor the code; tweak some network parameters * Simplify the segmentation network structure; make the object detection network more complex in turn * Problem: CUDA errors not reported properly to console Solution: stop and join data loader threads even in case of exceptions * Minor parameters tweaking * Loss may have increased, even if prob_loss_increasing_thresh > prob_loss_increasing_thresh_max_value * Add previous_loss_values_dump_amount to previous_loss_values.size() when deciding if loss has been increasing * Improve behaviour when loss actually increased after disk sync * Revert some of the earlier change * Disregard dumped loss values only when deciding if learning rate should be shrunk, but *not* when deciding if loss has been going up since last disk sync * Revert "Revert some of the earlier change" This reverts commit 6c852124efe6473a5c962de0091709129d6fcde3. * Keep enough previous loss values, until the disk sync * Fix maintaining the dumped (now "effectively disregarded") loss values count * Detect cats instead of aeroplanes * Add helpful logging * Clarify the intention and the code * Review fixes * Add operator== for the other pixel types as well; remove the inline * If available, use constexpr if * Revert "If available, use constexpr if" This reverts commit 503d4dd3355ff8ad613116e3ffcc0fa664674f69. * Simplify code as per review comments * Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh * Clarify console output * Revert "Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh" This reverts commit 9191ebc7762d17d81cdfc334a80ca9a667365740. * To keep the changes to a bare minimum, revert the steps_since_last_learning_rate_shrink change after all (at least for now) * Even empty out some of the previous test loss values * Minor review fixes * Can't use C++14 features here * Do not use the struct name as a variable name
2019-11-15 11:53:16 +08:00
// The contents of this file are in the public domain. See LICENSE_FOR_EXAMPLE_PROGRAMS.txt
/*
This example shows how to do instance segmentation on an image using net pretrained
on the PASCAL VOC2012 dataset. For an introduction to what instance segmentation is,
see the accompanying header file dnn_instance_segmentation_ex.h.
Instructions how to run the example:
1. Download the PASCAL VOC2012 data, and untar it somewhere.
http://host.robots.ox.ac.uk/pascal/VOC/voc2012/VOCtrainval_11-May-2012.tar
2. Build the dnn_instance_segmentation_train_ex example program.
3. Run:
./dnn_instance_segmentation_train_ex /path/to/VOC2012
4. Wait while the network is being trained.
5. Build the dnn_instance_segmentation_ex example program.
6. Run:
./dnn_instance_segmentation_ex /path/to/VOC2012-or-other-images
An alternative to steps 2-4 above is to download a pre-trained network
from here: http://dlib.net/files/instance_segmentation_voc2012net_v2.dnn
Instance segmentation (#1918) * Add instance segmentation example - first version of training code * Add MMOD options; get rid of the cache approach, and instead load all MMOD rects upfront * Improve console output * Set filter count * Minor tweaking * Inference - first version, at least compiles! * Ignore overlapped boxes * Ignore even small instances * Set overlaps_ignore * Add TODO remarks * Revert "Set overlaps_ignore" This reverts commit 65adeff1f89af62b10c691e7aa86c04fc358d03e. * Set result size * Set label image size * Take ignore-color into account * Fix the cropping rect's aspect ratio; also slightly expand the rect * Draw the largest findings last * Improve masking of the current instance * Add some perturbation to the inputs * Simplify ground-truth reading; fix random cropping * Read even class labels * Tweak default minibatch size * Learn only one class * Really train only instances of the selected class * Remove outdated TODO remark * Automatically skip images with no detections * Print to console what was found * Fix class index problem * Fix indentation * Allow to choose multiple classes * Draw rect in the color of the corresponding class * Write detector window classes to ostream; also group detection windows by class (when ostreaming) * Train a separate instance segmentation network for each classlabel * Use separate synchronization file for each seg net of each class * Allow more overlap * Fix sorting criterion * Fix interpolating the predicted mask * Improve bilinear interpolation: if output type is an integer, round instead of truncating * Add helpful comments * Ignore large aspect ratios; refactor the code; tweak some network parameters * Simplify the segmentation network structure; make the object detection network more complex in turn * Problem: CUDA errors not reported properly to console Solution: stop and join data loader threads even in case of exceptions * Minor parameters tweaking * Loss may have increased, even if prob_loss_increasing_thresh > prob_loss_increasing_thresh_max_value * Add previous_loss_values_dump_amount to previous_loss_values.size() when deciding if loss has been increasing * Improve behaviour when loss actually increased after disk sync * Revert some of the earlier change * Disregard dumped loss values only when deciding if learning rate should be shrunk, but *not* when deciding if loss has been going up since last disk sync * Revert "Revert some of the earlier change" This reverts commit 6c852124efe6473a5c962de0091709129d6fcde3. * Keep enough previous loss values, until the disk sync * Fix maintaining the dumped (now "effectively disregarded") loss values count * Detect cats instead of aeroplanes * Add helpful logging * Clarify the intention and the code * Review fixes * Add operator== for the other pixel types as well; remove the inline * If available, use constexpr if * Revert "If available, use constexpr if" This reverts commit 503d4dd3355ff8ad613116e3ffcc0fa664674f69. * Simplify code as per review comments * Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh * Clarify console output * Revert "Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh" This reverts commit 9191ebc7762d17d81cdfc334a80ca9a667365740. * To keep the changes to a bare minimum, revert the steps_since_last_learning_rate_shrink change after all (at least for now) * Even empty out some of the previous test loss values * Minor review fixes * Can't use C++14 features here * Do not use the struct name as a variable name
2019-11-15 11:53:16 +08:00
It would be a good idea to become familiar with dlib's DNN tooling before reading this
example. So you should read dnn_introduction_ex.cpp and dnn_introduction2_ex.cpp
before reading this example program.
*/
#include "dnn_instance_segmentation_ex.h"
#include "pascal_voc_2012.h"
#include <iostream>
#include <dlib/data_io.h>
#include <dlib/gui_widgets.h>
using namespace std;
using namespace dlib;
// ----------------------------------------------------------------------------------------
int main(int argc, char** argv) try
{
if (argc != 2)
{
cout << "You call this program like this: " << endl;
cout << "./dnn_instance_segmentation_train_ex /path/to/images" << endl;
cout << endl;
cout << "You will also need a trained '" << instance_segmentation_net_filename << "' file." << endl;
cout << "You can either train it yourself (see example program" << endl;
cout << "dnn_instance_segmentation_train_ex), or download a" << endl;
cout << "copy from here: http://dlib.net/files/" << instance_segmentation_net_filename << endl;
return 1;
}
// Read the file containing the trained networks from the working directory.
det_anet_type det_net;
std::map<std::string, seg_bnet_type> seg_nets_by_class;
deserialize(instance_segmentation_net_filename) >> det_net >> seg_nets_by_class;
// Show inference results in a window.
image_window win;
matrix<rgb_pixel> input_image;
// Find supported image files.
const std::vector<file> files = dlib::get_files_in_directory_tree(argv[1],
dlib::match_endings(".jpeg .jpg .png .webp"));
Instance segmentation (#1918) * Add instance segmentation example - first version of training code * Add MMOD options; get rid of the cache approach, and instead load all MMOD rects upfront * Improve console output * Set filter count * Minor tweaking * Inference - first version, at least compiles! * Ignore overlapped boxes * Ignore even small instances * Set overlaps_ignore * Add TODO remarks * Revert "Set overlaps_ignore" This reverts commit 65adeff1f89af62b10c691e7aa86c04fc358d03e. * Set result size * Set label image size * Take ignore-color into account * Fix the cropping rect's aspect ratio; also slightly expand the rect * Draw the largest findings last * Improve masking of the current instance * Add some perturbation to the inputs * Simplify ground-truth reading; fix random cropping * Read even class labels * Tweak default minibatch size * Learn only one class * Really train only instances of the selected class * Remove outdated TODO remark * Automatically skip images with no detections * Print to console what was found * Fix class index problem * Fix indentation * Allow to choose multiple classes * Draw rect in the color of the corresponding class * Write detector window classes to ostream; also group detection windows by class (when ostreaming) * Train a separate instance segmentation network for each classlabel * Use separate synchronization file for each seg net of each class * Allow more overlap * Fix sorting criterion * Fix interpolating the predicted mask * Improve bilinear interpolation: if output type is an integer, round instead of truncating * Add helpful comments * Ignore large aspect ratios; refactor the code; tweak some network parameters * Simplify the segmentation network structure; make the object detection network more complex in turn * Problem: CUDA errors not reported properly to console Solution: stop and join data loader threads even in case of exceptions * Minor parameters tweaking * Loss may have increased, even if prob_loss_increasing_thresh > prob_loss_increasing_thresh_max_value * Add previous_loss_values_dump_amount to previous_loss_values.size() when deciding if loss has been increasing * Improve behaviour when loss actually increased after disk sync * Revert some of the earlier change * Disregard dumped loss values only when deciding if learning rate should be shrunk, but *not* when deciding if loss has been going up since last disk sync * Revert "Revert some of the earlier change" This reverts commit 6c852124efe6473a5c962de0091709129d6fcde3. * Keep enough previous loss values, until the disk sync * Fix maintaining the dumped (now "effectively disregarded") loss values count * Detect cats instead of aeroplanes * Add helpful logging * Clarify the intention and the code * Review fixes * Add operator== for the other pixel types as well; remove the inline * If available, use constexpr if * Revert "If available, use constexpr if" This reverts commit 503d4dd3355ff8ad613116e3ffcc0fa664674f69. * Simplify code as per review comments * Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh * Clarify console output * Revert "Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh" This reverts commit 9191ebc7762d17d81cdfc334a80ca9a667365740. * To keep the changes to a bare minimum, revert the steps_since_last_learning_rate_shrink change after all (at least for now) * Even empty out some of the previous test loss values * Minor review fixes * Can't use C++14 features here * Do not use the struct name as a variable name
2019-11-15 11:53:16 +08:00
dlib::rand rnd;
cout << "Found " << files.size() << " images, processing..." << endl;
for (const file& file : files)
{
// Load the input image.
load_image(input_image, file.full_name());
Instance segmentation (#1918) * Add instance segmentation example - first version of training code * Add MMOD options; get rid of the cache approach, and instead load all MMOD rects upfront * Improve console output * Set filter count * Minor tweaking * Inference - first version, at least compiles! * Ignore overlapped boxes * Ignore even small instances * Set overlaps_ignore * Add TODO remarks * Revert "Set overlaps_ignore" This reverts commit 65adeff1f89af62b10c691e7aa86c04fc358d03e. * Set result size * Set label image size * Take ignore-color into account * Fix the cropping rect's aspect ratio; also slightly expand the rect * Draw the largest findings last * Improve masking of the current instance * Add some perturbation to the inputs * Simplify ground-truth reading; fix random cropping * Read even class labels * Tweak default minibatch size * Learn only one class * Really train only instances of the selected class * Remove outdated TODO remark * Automatically skip images with no detections * Print to console what was found * Fix class index problem * Fix indentation * Allow to choose multiple classes * Draw rect in the color of the corresponding class * Write detector window classes to ostream; also group detection windows by class (when ostreaming) * Train a separate instance segmentation network for each classlabel * Use separate synchronization file for each seg net of each class * Allow more overlap * Fix sorting criterion * Fix interpolating the predicted mask * Improve bilinear interpolation: if output type is an integer, round instead of truncating * Add helpful comments * Ignore large aspect ratios; refactor the code; tweak some network parameters * Simplify the segmentation network structure; make the object detection network more complex in turn * Problem: CUDA errors not reported properly to console Solution: stop and join data loader threads even in case of exceptions * Minor parameters tweaking * Loss may have increased, even if prob_loss_increasing_thresh > prob_loss_increasing_thresh_max_value * Add previous_loss_values_dump_amount to previous_loss_values.size() when deciding if loss has been increasing * Improve behaviour when loss actually increased after disk sync * Revert some of the earlier change * Disregard dumped loss values only when deciding if learning rate should be shrunk, but *not* when deciding if loss has been going up since last disk sync * Revert "Revert some of the earlier change" This reverts commit 6c852124efe6473a5c962de0091709129d6fcde3. * Keep enough previous loss values, until the disk sync * Fix maintaining the dumped (now "effectively disregarded") loss values count * Detect cats instead of aeroplanes * Add helpful logging * Clarify the intention and the code * Review fixes * Add operator== for the other pixel types as well; remove the inline * If available, use constexpr if * Revert "If available, use constexpr if" This reverts commit 503d4dd3355ff8ad613116e3ffcc0fa664674f69. * Simplify code as per review comments * Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh * Clarify console output * Revert "Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh" This reverts commit 9191ebc7762d17d81cdfc334a80ca9a667365740. * To keep the changes to a bare minimum, revert the steps_since_last_learning_rate_shrink change after all (at least for now) * Even empty out some of the previous test loss values * Minor review fixes * Can't use C++14 features here * Do not use the struct name as a variable name
2019-11-15 11:53:16 +08:00
// Find instances in the input image
const auto instances = det_net(input_image);
Instance segmentation (#1918) * Add instance segmentation example - first version of training code * Add MMOD options; get rid of the cache approach, and instead load all MMOD rects upfront * Improve console output * Set filter count * Minor tweaking * Inference - first version, at least compiles! * Ignore overlapped boxes * Ignore even small instances * Set overlaps_ignore * Add TODO remarks * Revert "Set overlaps_ignore" This reverts commit 65adeff1f89af62b10c691e7aa86c04fc358d03e. * Set result size * Set label image size * Take ignore-color into account * Fix the cropping rect's aspect ratio; also slightly expand the rect * Draw the largest findings last * Improve masking of the current instance * Add some perturbation to the inputs * Simplify ground-truth reading; fix random cropping * Read even class labels * Tweak default minibatch size * Learn only one class * Really train only instances of the selected class * Remove outdated TODO remark * Automatically skip images with no detections * Print to console what was found * Fix class index problem * Fix indentation * Allow to choose multiple classes * Draw rect in the color of the corresponding class * Write detector window classes to ostream; also group detection windows by class (when ostreaming) * Train a separate instance segmentation network for each classlabel * Use separate synchronization file for each seg net of each class * Allow more overlap * Fix sorting criterion * Fix interpolating the predicted mask * Improve bilinear interpolation: if output type is an integer, round instead of truncating * Add helpful comments * Ignore large aspect ratios; refactor the code; tweak some network parameters * Simplify the segmentation network structure; make the object detection network more complex in turn * Problem: CUDA errors not reported properly to console Solution: stop and join data loader threads even in case of exceptions * Minor parameters tweaking * Loss may have increased, even if prob_loss_increasing_thresh > prob_loss_increasing_thresh_max_value * Add previous_loss_values_dump_amount to previous_loss_values.size() when deciding if loss has been increasing * Improve behaviour when loss actually increased after disk sync * Revert some of the earlier change * Disregard dumped loss values only when deciding if learning rate should be shrunk, but *not* when deciding if loss has been going up since last disk sync * Revert "Revert some of the earlier change" This reverts commit 6c852124efe6473a5c962de0091709129d6fcde3. * Keep enough previous loss values, until the disk sync * Fix maintaining the dumped (now "effectively disregarded") loss values count * Detect cats instead of aeroplanes * Add helpful logging * Clarify the intention and the code * Review fixes * Add operator== for the other pixel types as well; remove the inline * If available, use constexpr if * Revert "If available, use constexpr if" This reverts commit 503d4dd3355ff8ad613116e3ffcc0fa664674f69. * Simplify code as per review comments * Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh * Clarify console output * Revert "Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh" This reverts commit 9191ebc7762d17d81cdfc334a80ca9a667365740. * To keep the changes to a bare minimum, revert the steps_since_last_learning_rate_shrink change after all (at least for now) * Even empty out some of the previous test loss values * Minor review fixes * Can't use C++14 features here * Do not use the struct name as a variable name
2019-11-15 11:53:16 +08:00
matrix<rgb_pixel> rgb_label_image;
matrix<float> label_image_confidence;
Instance segmentation (#1918) * Add instance segmentation example - first version of training code * Add MMOD options; get rid of the cache approach, and instead load all MMOD rects upfront * Improve console output * Set filter count * Minor tweaking * Inference - first version, at least compiles! * Ignore overlapped boxes * Ignore even small instances * Set overlaps_ignore * Add TODO remarks * Revert "Set overlaps_ignore" This reverts commit 65adeff1f89af62b10c691e7aa86c04fc358d03e. * Set result size * Set label image size * Take ignore-color into account * Fix the cropping rect's aspect ratio; also slightly expand the rect * Draw the largest findings last * Improve masking of the current instance * Add some perturbation to the inputs * Simplify ground-truth reading; fix random cropping * Read even class labels * Tweak default minibatch size * Learn only one class * Really train only instances of the selected class * Remove outdated TODO remark * Automatically skip images with no detections * Print to console what was found * Fix class index problem * Fix indentation * Allow to choose multiple classes * Draw rect in the color of the corresponding class * Write detector window classes to ostream; also group detection windows by class (when ostreaming) * Train a separate instance segmentation network for each classlabel * Use separate synchronization file for each seg net of each class * Allow more overlap * Fix sorting criterion * Fix interpolating the predicted mask * Improve bilinear interpolation: if output type is an integer, round instead of truncating * Add helpful comments * Ignore large aspect ratios; refactor the code; tweak some network parameters * Simplify the segmentation network structure; make the object detection network more complex in turn * Problem: CUDA errors not reported properly to console Solution: stop and join data loader threads even in case of exceptions * Minor parameters tweaking * Loss may have increased, even if prob_loss_increasing_thresh > prob_loss_increasing_thresh_max_value * Add previous_loss_values_dump_amount to previous_loss_values.size() when deciding if loss has been increasing * Improve behaviour when loss actually increased after disk sync * Revert some of the earlier change * Disregard dumped loss values only when deciding if learning rate should be shrunk, but *not* when deciding if loss has been going up since last disk sync * Revert "Revert some of the earlier change" This reverts commit 6c852124efe6473a5c962de0091709129d6fcde3. * Keep enough previous loss values, until the disk sync * Fix maintaining the dumped (now "effectively disregarded") loss values count * Detect cats instead of aeroplanes * Add helpful logging * Clarify the intention and the code * Review fixes * Add operator== for the other pixel types as well; remove the inline * If available, use constexpr if * Revert "If available, use constexpr if" This reverts commit 503d4dd3355ff8ad613116e3ffcc0fa664674f69. * Simplify code as per review comments * Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh * Clarify console output * Revert "Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh" This reverts commit 9191ebc7762d17d81cdfc334a80ca9a667365740. * To keep the changes to a bare minimum, revert the steps_since_last_learning_rate_shrink change after all (at least for now) * Even empty out some of the previous test loss values * Minor review fixes * Can't use C++14 features here * Do not use the struct name as a variable name
2019-11-15 11:53:16 +08:00
matrix<rgb_pixel> input_chip;
rgb_label_image.set_size(input_image.nr(), input_image.nc());
rgb_label_image = rgb_pixel(0, 0, 0);
label_image_confidence.set_size(input_image.nr(), input_image.nc());
label_image_confidence = 0.0;
Instance segmentation (#1918) * Add instance segmentation example - first version of training code * Add MMOD options; get rid of the cache approach, and instead load all MMOD rects upfront * Improve console output * Set filter count * Minor tweaking * Inference - first version, at least compiles! * Ignore overlapped boxes * Ignore even small instances * Set overlaps_ignore * Add TODO remarks * Revert "Set overlaps_ignore" This reverts commit 65adeff1f89af62b10c691e7aa86c04fc358d03e. * Set result size * Set label image size * Take ignore-color into account * Fix the cropping rect's aspect ratio; also slightly expand the rect * Draw the largest findings last * Improve masking of the current instance * Add some perturbation to the inputs * Simplify ground-truth reading; fix random cropping * Read even class labels * Tweak default minibatch size * Learn only one class * Really train only instances of the selected class * Remove outdated TODO remark * Automatically skip images with no detections * Print to console what was found * Fix class index problem * Fix indentation * Allow to choose multiple classes * Draw rect in the color of the corresponding class * Write detector window classes to ostream; also group detection windows by class (when ostreaming) * Train a separate instance segmentation network for each classlabel * Use separate synchronization file for each seg net of each class * Allow more overlap * Fix sorting criterion * Fix interpolating the predicted mask * Improve bilinear interpolation: if output type is an integer, round instead of truncating * Add helpful comments * Ignore large aspect ratios; refactor the code; tweak some network parameters * Simplify the segmentation network structure; make the object detection network more complex in turn * Problem: CUDA errors not reported properly to console Solution: stop and join data loader threads even in case of exceptions * Minor parameters tweaking * Loss may have increased, even if prob_loss_increasing_thresh > prob_loss_increasing_thresh_max_value * Add previous_loss_values_dump_amount to previous_loss_values.size() when deciding if loss has been increasing * Improve behaviour when loss actually increased after disk sync * Revert some of the earlier change * Disregard dumped loss values only when deciding if learning rate should be shrunk, but *not* when deciding if loss has been going up since last disk sync * Revert "Revert some of the earlier change" This reverts commit 6c852124efe6473a5c962de0091709129d6fcde3. * Keep enough previous loss values, until the disk sync * Fix maintaining the dumped (now "effectively disregarded") loss values count * Detect cats instead of aeroplanes * Add helpful logging * Clarify the intention and the code * Review fixes * Add operator== for the other pixel types as well; remove the inline * If available, use constexpr if * Revert "If available, use constexpr if" This reverts commit 503d4dd3355ff8ad613116e3ffcc0fa664674f69. * Simplify code as per review comments * Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh * Clarify console output * Revert "Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh" This reverts commit 9191ebc7762d17d81cdfc334a80ca9a667365740. * To keep the changes to a bare minimum, revert the steps_since_last_learning_rate_shrink change after all (at least for now) * Even empty out some of the previous test loss values * Minor review fixes * Can't use C++14 features here * Do not use the struct name as a variable name
2019-11-15 11:53:16 +08:00
bool found_something = false;
for (const auto& instance : instances)
{
if (!found_something)
{
cout << "Found ";
found_something = true;
}
else
{
cout << ", ";
}
cout << instance.label;
const auto cropping_rect = get_cropping_rect(instance.rect);
const chip_details chip_details(cropping_rect, chip_dims(seg_dim, seg_dim));
extract_image_chip(input_image, chip_details, input_chip, interpolate_bilinear());
const auto i = seg_nets_by_class.find(instance.label);
if (i == seg_nets_by_class.end())
{
// per-class segmentation net not found, so we must be using the same net for all classes
// (see bool separate_seg_net_for_each_class in dnn_instance_segmentation_train_ex.cpp)
DLIB_CASSERT(seg_nets_by_class.size() == 1);
DLIB_CASSERT(seg_nets_by_class.begin()->first == "");
}
auto& seg_net = i != seg_nets_by_class.end()
? i->second // use the segmentation net trained for this class
: seg_nets_by_class.begin()->second; // use the same segmentation net for all classes
const auto mask = seg_net(input_chip);
const rgb_pixel random_color(
rnd.get_random_8bit_number(),
rnd.get_random_8bit_number(),
rnd.get_random_8bit_number()
);
dlib::matrix<float> resized_mask(
Instance segmentation (#1918) * Add instance segmentation example - first version of training code * Add MMOD options; get rid of the cache approach, and instead load all MMOD rects upfront * Improve console output * Set filter count * Minor tweaking * Inference - first version, at least compiles! * Ignore overlapped boxes * Ignore even small instances * Set overlaps_ignore * Add TODO remarks * Revert "Set overlaps_ignore" This reverts commit 65adeff1f89af62b10c691e7aa86c04fc358d03e. * Set result size * Set label image size * Take ignore-color into account * Fix the cropping rect's aspect ratio; also slightly expand the rect * Draw the largest findings last * Improve masking of the current instance * Add some perturbation to the inputs * Simplify ground-truth reading; fix random cropping * Read even class labels * Tweak default minibatch size * Learn only one class * Really train only instances of the selected class * Remove outdated TODO remark * Automatically skip images with no detections * Print to console what was found * Fix class index problem * Fix indentation * Allow to choose multiple classes * Draw rect in the color of the corresponding class * Write detector window classes to ostream; also group detection windows by class (when ostreaming) * Train a separate instance segmentation network for each classlabel * Use separate synchronization file for each seg net of each class * Allow more overlap * Fix sorting criterion * Fix interpolating the predicted mask * Improve bilinear interpolation: if output type is an integer, round instead of truncating * Add helpful comments * Ignore large aspect ratios; refactor the code; tweak some network parameters * Simplify the segmentation network structure; make the object detection network more complex in turn * Problem: CUDA errors not reported properly to console Solution: stop and join data loader threads even in case of exceptions * Minor parameters tweaking * Loss may have increased, even if prob_loss_increasing_thresh > prob_loss_increasing_thresh_max_value * Add previous_loss_values_dump_amount to previous_loss_values.size() when deciding if loss has been increasing * Improve behaviour when loss actually increased after disk sync * Revert some of the earlier change * Disregard dumped loss values only when deciding if learning rate should be shrunk, but *not* when deciding if loss has been going up since last disk sync * Revert "Revert some of the earlier change" This reverts commit 6c852124efe6473a5c962de0091709129d6fcde3. * Keep enough previous loss values, until the disk sync * Fix maintaining the dumped (now "effectively disregarded") loss values count * Detect cats instead of aeroplanes * Add helpful logging * Clarify the intention and the code * Review fixes * Add operator== for the other pixel types as well; remove the inline * If available, use constexpr if * Revert "If available, use constexpr if" This reverts commit 503d4dd3355ff8ad613116e3ffcc0fa664674f69. * Simplify code as per review comments * Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh * Clarify console output * Revert "Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh" This reverts commit 9191ebc7762d17d81cdfc334a80ca9a667365740. * To keep the changes to a bare minimum, revert the steps_since_last_learning_rate_shrink change after all (at least for now) * Even empty out some of the previous test loss values * Minor review fixes * Can't use C++14 features here * Do not use the struct name as a variable name
2019-11-15 11:53:16 +08:00
static_cast<int>(chip_details.rect.height()),
static_cast<int>(chip_details.rect.width())
);
dlib::resize_image(mask, resized_mask);
for (int r = 0; r < resized_mask.nr(); ++r)
{
for (int c = 0; c < resized_mask.nc(); ++c)
{
const auto new_confidence = resized_mask(r, c);
if (new_confidence > 0)
Instance segmentation (#1918) * Add instance segmentation example - first version of training code * Add MMOD options; get rid of the cache approach, and instead load all MMOD rects upfront * Improve console output * Set filter count * Minor tweaking * Inference - first version, at least compiles! * Ignore overlapped boxes * Ignore even small instances * Set overlaps_ignore * Add TODO remarks * Revert "Set overlaps_ignore" This reverts commit 65adeff1f89af62b10c691e7aa86c04fc358d03e. * Set result size * Set label image size * Take ignore-color into account * Fix the cropping rect's aspect ratio; also slightly expand the rect * Draw the largest findings last * Improve masking of the current instance * Add some perturbation to the inputs * Simplify ground-truth reading; fix random cropping * Read even class labels * Tweak default minibatch size * Learn only one class * Really train only instances of the selected class * Remove outdated TODO remark * Automatically skip images with no detections * Print to console what was found * Fix class index problem * Fix indentation * Allow to choose multiple classes * Draw rect in the color of the corresponding class * Write detector window classes to ostream; also group detection windows by class (when ostreaming) * Train a separate instance segmentation network for each classlabel * Use separate synchronization file for each seg net of each class * Allow more overlap * Fix sorting criterion * Fix interpolating the predicted mask * Improve bilinear interpolation: if output type is an integer, round instead of truncating * Add helpful comments * Ignore large aspect ratios; refactor the code; tweak some network parameters * Simplify the segmentation network structure; make the object detection network more complex in turn * Problem: CUDA errors not reported properly to console Solution: stop and join data loader threads even in case of exceptions * Minor parameters tweaking * Loss may have increased, even if prob_loss_increasing_thresh > prob_loss_increasing_thresh_max_value * Add previous_loss_values_dump_amount to previous_loss_values.size() when deciding if loss has been increasing * Improve behaviour when loss actually increased after disk sync * Revert some of the earlier change * Disregard dumped loss values only when deciding if learning rate should be shrunk, but *not* when deciding if loss has been going up since last disk sync * Revert "Revert some of the earlier change" This reverts commit 6c852124efe6473a5c962de0091709129d6fcde3. * Keep enough previous loss values, until the disk sync * Fix maintaining the dumped (now "effectively disregarded") loss values count * Detect cats instead of aeroplanes * Add helpful logging * Clarify the intention and the code * Review fixes * Add operator== for the other pixel types as well; remove the inline * If available, use constexpr if * Revert "If available, use constexpr if" This reverts commit 503d4dd3355ff8ad613116e3ffcc0fa664674f69. * Simplify code as per review comments * Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh * Clarify console output * Revert "Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh" This reverts commit 9191ebc7762d17d81cdfc334a80ca9a667365740. * To keep the changes to a bare minimum, revert the steps_since_last_learning_rate_shrink change after all (at least for now) * Even empty out some of the previous test loss values * Minor review fixes * Can't use C++14 features here * Do not use the struct name as a variable name
2019-11-15 11:53:16 +08:00
{
const auto y = chip_details.rect.top() + r;
const auto x = chip_details.rect.left() + c;
if (y >= 0 && y < rgb_label_image.nr() && x >= 0 && x < rgb_label_image.nc())
{
auto& current_confidence = label_image_confidence(y, x);
if (new_confidence > current_confidence)
{
auto rgb_label = random_color;
const auto baseline_confidence = 5;
if (new_confidence < baseline_confidence)
{
// Scale label intensity if confidence isn't high
rgb_label.red *= new_confidence / baseline_confidence;
rgb_label.green *= new_confidence / baseline_confidence;
rgb_label.blue *= new_confidence / baseline_confidence;
}
rgb_label_image(y, x) = rgb_label;
current_confidence = new_confidence;
}
}
Instance segmentation (#1918) * Add instance segmentation example - first version of training code * Add MMOD options; get rid of the cache approach, and instead load all MMOD rects upfront * Improve console output * Set filter count * Minor tweaking * Inference - first version, at least compiles! * Ignore overlapped boxes * Ignore even small instances * Set overlaps_ignore * Add TODO remarks * Revert "Set overlaps_ignore" This reverts commit 65adeff1f89af62b10c691e7aa86c04fc358d03e. * Set result size * Set label image size * Take ignore-color into account * Fix the cropping rect's aspect ratio; also slightly expand the rect * Draw the largest findings last * Improve masking of the current instance * Add some perturbation to the inputs * Simplify ground-truth reading; fix random cropping * Read even class labels * Tweak default minibatch size * Learn only one class * Really train only instances of the selected class * Remove outdated TODO remark * Automatically skip images with no detections * Print to console what was found * Fix class index problem * Fix indentation * Allow to choose multiple classes * Draw rect in the color of the corresponding class * Write detector window classes to ostream; also group detection windows by class (when ostreaming) * Train a separate instance segmentation network for each classlabel * Use separate synchronization file for each seg net of each class * Allow more overlap * Fix sorting criterion * Fix interpolating the predicted mask * Improve bilinear interpolation: if output type is an integer, round instead of truncating * Add helpful comments * Ignore large aspect ratios; refactor the code; tweak some network parameters * Simplify the segmentation network structure; make the object detection network more complex in turn * Problem: CUDA errors not reported properly to console Solution: stop and join data loader threads even in case of exceptions * Minor parameters tweaking * Loss may have increased, even if prob_loss_increasing_thresh > prob_loss_increasing_thresh_max_value * Add previous_loss_values_dump_amount to previous_loss_values.size() when deciding if loss has been increasing * Improve behaviour when loss actually increased after disk sync * Revert some of the earlier change * Disregard dumped loss values only when deciding if learning rate should be shrunk, but *not* when deciding if loss has been going up since last disk sync * Revert "Revert some of the earlier change" This reverts commit 6c852124efe6473a5c962de0091709129d6fcde3. * Keep enough previous loss values, until the disk sync * Fix maintaining the dumped (now "effectively disregarded") loss values count * Detect cats instead of aeroplanes * Add helpful logging * Clarify the intention and the code * Review fixes * Add operator== for the other pixel types as well; remove the inline * If available, use constexpr if * Revert "If available, use constexpr if" This reverts commit 503d4dd3355ff8ad613116e3ffcc0fa664674f69. * Simplify code as per review comments * Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh * Clarify console output * Revert "Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh" This reverts commit 9191ebc7762d17d81cdfc334a80ca9a667365740. * To keep the changes to a bare minimum, revert the steps_since_last_learning_rate_shrink change after all (at least for now) * Even empty out some of the previous test loss values * Minor review fixes * Can't use C++14 features here * Do not use the struct name as a variable name
2019-11-15 11:53:16 +08:00
}
}
}
const Voc2012class& voc2012_class = find_voc2012_class(
[&instance](const Voc2012class& candidate) {
return candidate.classlabel == instance.label;
}
);
dlib::draw_rectangle(rgb_label_image, instance.rect, voc2012_class.rgb_label, 1);
}
// Show the input image on the left, and the predicted RGB labels on the right.
win.set_image(join_rows(input_image, rgb_label_image));
if (!instances.empty())
{
cout << " in " << file.name() << " - hit enter to process the next image";
cin.get();
}
}
}
catch(std::exception& e)
{
cout << e.what() << endl;
}