From 4aecaae94cf59148d6cc09301803328f2c5321ca Mon Sep 17 00:00:00 2001
From: MaxJa4 <74194322+MaxJa4@users.noreply.github.com>
Date: Sun, 12 Nov 2023 16:10:53 +0100
Subject: [PATCH 1/3] Update Readme.md

---
 doc/03_research/02_perception/Readme.md | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/doc/03_research/02_perception/Readme.md b/doc/03_research/02_perception/Readme.md
index 36d22f9f..157efa2d 100644
--- a/doc/03_research/02_perception/Readme.md
+++ b/doc/03_research/02_perception/Readme.md
@@ -1,5 +1,9 @@
 # Perception
 
-This folder contains all the results of our research on perception:
+This folder contains all the results of research on perception:
 
-* [Basics](./02_basics.md)
+* **PAF22**
+  * [Basics](./02_basics.md)
+  * [First implementation plan](./03_first_implementation_plan.md)
+* **PAF23**
+  * [Pylot](./04_pylot.md)

From 73b26a0eab19a65b06c4e4e8f842fa928dc3f37f Mon Sep 17 00:00:00 2001
From: MaxJa4 <74194322+MaxJa4@users.noreply.github.com>
Date: Sun, 12 Nov 2023 16:11:24 +0100
Subject: [PATCH 2/3] Create pylot docs

---
 doc/03_research/02_perception/04_pylot.md | 54 +++++++++++++++++++++++
 1 file changed, 54 insertions(+)
 create mode 100644 doc/03_research/02_perception/04_pylot.md

diff --git a/doc/03_research/02_perception/04_pylot.md b/doc/03_research/02_perception/04_pylot.md
new file mode 100644
index 00000000..6673f9c2
--- /dev/null
+++ b/doc/03_research/02_perception/04_pylot.md
@@ -0,0 +1,54 @@
+# Pylot - Perception
+
+**Authors:** Maximilian Jannack
+
+**Date:** 2.11.2023
+
+---
+
+## [Detection](https://pylot.readthedocs.io/en/latest/perception.detection.html)
+
+### Obstacle detection
+
+Pylot provides two options for obstacle detection:
+
+1. Obstacle detection operator that can use any model that adheres to the Tensorflow `object detection model zoo`
+   - By default, three models that were trained on 1080p CARLA images (`faster-rcnn`, `ssd-mobilenet-fpn-640`, and `ssdlit-mobilenet-v2`) are provided
+   - Models that have been trained on other data sets can be easily plugged in
+2. Operator that can infer any of the EfficientDet models (not trained on CARLA data, but on the COCO dataset)
+
+### Traffic light detection
+
+Uses `Faster RCNN weight` (trained on 1080p CARLA images)
+
+### Lane detection
+
+Uses the `Lanenet` model ([repo](https://github.com/MaybeShewill-CV/lanenet-lane-detection)) or canny edge detector.
+
+---
+
+## [Obstacle Tracking](https://pylot.readthedocs.io/en/latest/perception.tracking.html)
+
+For tracking obstacles across frames.
+Uses the `DaSiamRPN` model ([repo](https://github.com/foolwood/DaSiamRPN)) to serially track multiple obstacles.
+
+---
+
+## [Depth Estimation](https://pylot.readthedocs.io/en/latest/perception.depth_estimation.html)
+
+Uses stereo cameras to estimate depth with the `AnyNet` model ([repo](https://github.com/mileyan/AnyNet)).
+Configurable camera distance between left and right.
+
+---
+
+## [Segmentation](https://pylot.readthedocs.io/en/latest/perception.segmentation.html)
+
+Different approaches, such as using the `DRN` model ([repo](https://github.com/ICGog/drn)) for segmenting camera images.
+No model with training on CARLA data available, output of segmentation component not used in Pylot right now.
+
+---
+
+## [Lidar](https://github.com/erdos-project/pylot/blob/master/pylot/perception/point_cloud.py)
+
+Pylot contains a few helpful function for handling point clouds from the Lidar sensor.
+It can for example merge two point clouds or map the points to a camera image.

From ca1224116ca6bba9c34e5a5ff45bb3e9ca4c7327 Mon Sep 17 00:00:00 2001
From: MaxJa4 <74194322+MaxJa4@users.noreply.github.com>
Date: Sun, 12 Nov 2023 16:31:07 +0100
Subject: [PATCH 3/3] Correct date

---
 doc/03_research/02_perception/04_pylot.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/doc/03_research/02_perception/04_pylot.md b/doc/03_research/02_perception/04_pylot.md
index 6673f9c2..3b82e29e 100644
--- a/doc/03_research/02_perception/04_pylot.md
+++ b/doc/03_research/02_perception/04_pylot.md
@@ -2,7 +2,7 @@
 
 **Authors:** Maximilian Jannack
 
-**Date:** 2.11.2023
+**Date:** 12.11.2023
 
 ---