Finish the project plan

doc/project-plan/sections/concept.tex

@@ -1,15 +1,34 @@
 \chapter{Concept}
 \label{ch:concept}
 
+This chapter briefly explains the concept of this project.
+The three main parts are the demonstrator, the training and the inference.
+
+\section{Demonstrator}
+\label{sec:demonstrator}
+
+The demonstrator --- a throwing booth --- was not entirely finished during the previous project.
+However, the things that still need to be done are not that time-consuming.
+For this reason, those tasks are not included in the timetable.
+They are carried out on the side to fill time gaps (e.g. during the training of a CNN model).
 
 \section{Training}
 \label{sec:training}
 
+The development and training of the CNN model is an iterative process.
+Various CNN architectures are experimented with and the results are compared to each other.
+Furthermore, the labeled dataset allows for some experimentation as well.
+For instance, each image has a label that indicates whether the object was fully or only partially visible.
 
 \section{Inference}
 \label{sec:inference}
 
+A high-performance solution requires fast image acquisition and low-latency inference.
+To implement the necessary components on the MPSoC with the integrated FPGA, a variety of toolchains and Linux-based operating systems can be used.
+These options are initially examined carefully in order to select the most suitable one.
 
-\section{Throwing Booth}
-\label{sec:throwing_booth}
+The specific implementation of the image acquisition and the inference depends on the overall performance and the available resources of the MPSoC development board.
+For this reason, several performance tests are carried out to evaluate its real-time capabilities.
 
+The final step is to verify the performance and accuracy of the entire system.
+The accuracy is additionally compared to the accuracy of the computer solution.

doc/project-plan/sections/deliverables.tex

@@ -10,8 +10,7 @@ \chapter{Deliverables}
   Project Plan & March 06, 2020 \\
   Demonstrator & August 14, 2020 \\
   Working Repository & August 14, 2020 \\
-  % Bachelor Thesis & August 14, 2020 \\
-  Bachelor Thesis incl. Developer Guide & August 14, 2020 \\
+  Bachelor Thesis & August 14, 2020 \\
   Poster & August 14, 2020 \\
   Thesis Defense & September 10/11, 2020 \\
   Fact Sheet & September 19, 2020 \\

doc/project-plan/sections/initial_situation.tex

@@ -7,5 +7,5 @@ \chapter{Initial Situation}
 The dataset is fully labeled and consists of more than \num{15000} usable images with at least 480 images of each object.
 
 In this project, a convolutional neural network (CNN) model is developed and trained on a computer using the collected dataset.
-This model is then implemented on an FPGA development board for a high-performance solution.
+This model is then implemented on an MPSoC development board for a high-performance solution.
 Since the throwing booth is used for demonstration purposes, the verification is an essential aspect of this project.

doc/project-plan/sections/project_goal.tex

@@ -1,7 +1,7 @@
 \chapter{Project Goal}
 \label{ch:project_goal}
 
-The goal of this project is to develop and train a CNN model and implement it on an FPGA development board.
+The goal of this project is to develop and train a CNN model and implement it on an MPSoC development board.
 Images of the throwing objects can then be captured by the camera, classified on the FPGA and displayed on the monitor.
 
 The respective objectives for this project are specified in the following table:\\