You can choose between three tasks:

Create a neural network to recognize your face and clearly discern it from other faces and objects

Create a chatbot. This chatbot shall use neural networks $($word$2$vec, LSTM$)$ for natural language processing of the human's textual input. As an output, it should present the requested information $/$ or hand over to a human operator. It is important to describe how this chatbot can be adapted to new tasks.

Own proposal $->$ must be handed in to Prof. Stache by $16\mathrm{.}06\mathrm{.}2024.$ Your proposal will only be accepted if it is on a similar or higher level of complexity compared to position $1$ or $2.$

For task $1,$ it is mandatory that you create own datasets by yourself; for tasks $2$ and $3$ it is preferred to do so$.$ However, you can complement your dataset by existing datasets from the internet. You must show examples of self$-$created data and state how many samples $($absolute and relative number$)$ you created by yourself.

For task $1$: You shall use transfer learning techniques. You must not use the VGG network as it is already available in the notebooks of the lecture

More precise requirements

Provide result in a jupyter notebook $(+$ external files for the data$)$

Notebook shall contain at least these sections:

Data preparation $($note: you need to create data$)$

Network training

$($requirements: use other network than VGG$,$ use transfer learning$)$

Evaluation

Discussion of the results $($also show examples where it fails, discuss why$)$

Notebook must be fully executable!

Notebook shall contain explanations in Markdown cells, clarifying what is done and why.

Language can be English or German

Results must be individual work, no group work, results are checked for plagiarism!

The notebook must be additionally handed in as pdf file $($use print to pdf of your browser$)$

Hand it in by upload to llias, due date: $11\mathrm{.}07\mathrm{.}2024,23$:$55.$

Put all necessary files into a $.$zip file name it by lastname$\_$firstname.zip

Other compressed formats such as $.$rar or $\mathrm{.}7$zip are not accepted $-$ work in such formats will not be graded!

Please also note that the upload is limited to $300$ MB $($via ILIAS$).$ Your work, including database must fit into this space.

For grading, refer to the grading policy in ILIAS

Project survey: After the submission deadline for your project work, a survey with detailed questions about your work will be activated. This survey is an important basis for evaluating the project and providing evidence of your own work performance. Therefore, please note that participation is mandatory and will be subject to time restrictions regarding the maximum processing time. The survey must be completed by July $14,11$:$55$ p$.$m$.$ at the latest, otherwise points will be deducted for your project work $($for details, see assessment guidelines$).$

AS PER ABOVE REQUIREMENTS, PLEASE SOLVE CORRECT, I WILL GIVE YOU GOOD REVIEW $!$