merge pifpaf - master (#357)

* feat: enable pifpaf inference support * refact: rm comments * feat: openpifpaf decoder finalized * fix: ci * fix: license badge * fix: gcc7 compilation * fix: conflicts w/ master branch ++

Author: ganler

README.md

@@ -85,8 +85,6 @@ We compare the prediction performance of HyperPose with [OpenPose 1.6](https://g
 | OpenPose (MobileNet) | 17.9 MB | 432 x 368 | **84.32 FPS** | 8.5 FPS (TF-Pose) |
 | OpenPose (ResNet18) | 45.0 MB | 432 x 368 | **62.52 FPS** | N/A |
 | OpenPifPaf (ResNet50) | 97.6 MB | 97 x 129 | **178.6 FPS** | 35.3 |
-<<<<<<< HEAD
-=======
 
 ## Accuracy
 We evaluate accuracy of pose estimation models developed by hyperpose (mainly over Mscoco2017 dataset). the development environment is Ubuntu16.04, with 4 V100-DGXs and 24 Intel Xeon CPU. The training procedure takes 1~2 weeks using 1 V100-DGX for each model. (If you want to train from strach, loading the pretrained backbone weight is recommended.)
@@ -99,7 +97,7 @@ We evaluate accuracy of pose estimation models developed by hyperpose (mainly ov
 | LightweightOpenpose (tinyvgg) | 23.6 MB | 432 x 368 | Mscoco2017(all 5000 images) | 47.3 map | - |
 | LightweightOpenpose (resnet50) | 42.7 MB | 432 x 368 | Mscoco2017(all 5000 images) | 48.2 map | - |
 | PoseProposal (resnet18) | 45.2 MB | 384 x 384 | MPII(all 2729 images) | 54.9 map (Pckh) | 72.8 map (Pckh)|
->>>>>>> master
+
 
 </a>
 <p align="center">

src/pifpaf_decoder/openpifpaf_postprocessor.cpp

@@ -461,11 +461,7 @@ namespace aiapp_impl {
  // frontierActive = true;
  // blockFrontier.clear();
  std::set<std::pair<int, int>> in_frontier{};
-<<<<<<< HEAD
  std::priority_queue<queue_item, std::deque<queue_item>, std::greater<>> frontier;
-=======
- std::priority_queue<queue_item, std::deque<queue_item>, std::greater<queue_item>> frontier;
->>>>>>> master
 
  const auto add_to_frontier = [&](size_t start_i) {
  for (const auto& [end_i, to_p] : BY_SOURCE_MAP[start_i]) {