py7dtd - A 7 Days to Die tools collection

https://7daystodie.com/

In this repository are collected tools and scripts for the game 7 Days to Die:

• Tools for the detection of objects/entities
• Scripts for the automatization of actions (mining, crafting, etc.)
• Aim bots
• Passcode cracking
• Block detection

Installing

This software can be installed as a pip module.

Windows

py -m pip install py7dtd

Unix

python -m pip install py7dtd

Usage

There are currently three functions available:

• Auto shooting
• Crack passcode
• Blocks detection

---

Contributing

Clone the module, create a virtual environment and install it:

git clone --recurse-submodules git@github.com:tassoneroberto/py7dtd.git
cd py7dtd
py -3.10 -m venv venv
.\venv\Scripts\Activate.ps1
py -m pip install .\iocontroller
py -m pip install .[ai]
ltt install torch torchvision

Note: if you are not interested in "entities detection" or "aimbot" then you can omit [ai] from the above command and the ltt install torch torchvision command:

py -m pip install .\iocontroller
py -m pip install .

Development mode

To install the package in edit mode (for developers) specify -e:

py -m pip install -e .\iocontroller
py -m pip install -e .[ai]
ltt install torch torchvision

Dependencies for entities detection

Disclaimer: an NVIDIA® GPU card with CUDA® architectures 3.5, 3.7, 5.2, 6.0, 6.1, 7.0 or higher is required. See the list of CUDA®-enabled GPU cards (https://developer.nvidia.com/cuda-gpus).

Install the following dependencies:

• C++ redistributable (https://support.microsoft.com/en-us/help/2977003/the-latest-supported-visual-c-downloads)
• NVIDIA® GPU drivers —CUDA® 11.2.1 (https://www.nvidia.com/drivers)
• CUDA® Toolkit 11.1 (https://developer.nvidia.com/cuda-toolkit-archive)
• cuDNN SDK 7.6 (https://developer.nvidia.com/cudnn)

In order to use the module you need a x64 version of Python 3.10.x. You can download it at this page: https://www.python.org/downloads/windows/

You also need to have a trained model for the entities detection in ./dataset/models/ and the dataset config json to in ./dataset/json/. You can train it yourself by following the instructions below or you can download it here:

• https://github.com/tassoneroberto/py7dtd/releases/download/v1.0.0-alpha1/yolov3_dataset_last.pt (copy to ./dataset/models/)
• https://github.com/tassoneroberto/py7dtd/releases/download/v1.0.0-alpha1/dataset_yolov3_detection_config.json (copy to ./dataset/json/)

https://github.com/tassoneroberto/py7dtd/releases/download/v1.0.0-alpha1/dataset_yolov3_detection_config.json

Training the entity detection model

The dataset used to train the model has to be increased to to improve the reliability of the detection. This can be done using the open source tool labelImg https://github.com/tzutalin/labelImg. The folder dataset needs to have the following structure:

• train
  • annotations
    • classes.txt
    • source1.txt
    • source2.txt
    • source3.txt
    • ...
  • images
    • source1.png
    • source2.png
    • source3.png
    • ...
• validation
  • annotations
    • classes.txt
    • source101.txt
    • source102.txt
    • source103.txt
    • ...
  • images
    • source101.png
    • source102.png
    • source103.png
    • ...

Run the following command to start training the model:

py7dtd_model_training

The following table is listing all the arguments that can be specified:

arg	description	default
help	Arguments description	N/A
dataset	Dataset folder path	./dataset
pretrained	Pre-trained model path	./dataset/models/yolov3.pt
epochs	Number of epochs to train	200

Notes

• The annotation has to be in the YOLO format.
• The test set has to be a lot smaller than the size of the train set.
• The classes.txt are automatically generated by labelImg and contain the names of the objects to detect.
• You can use a pre-trained model for transfer learning. To do so you have to download the file https://github.com/tassoneroberto/py7dtd/releases/download/v1.0.0-alpha1/yolov3.pt and put it in ./dataset/models (or specify a different location with the argument --pretrained).

Entities detection

❗ Under development ❗

Simple entities detector using AI (Computer Vision).

The objects detection (trees, zombies, etc.) is done using ImageAI: https://github.com/OlafenwaMoses/ImageAI/

The annotation of the images has been done using the tool labelImg: https://github.com/heartexlabs/labelImg/

Proof of concept

---

Aim bot

❗ Under development ❗

Simple aim bot capable of:

• Detect zombies/players
• Move the mouse to the target
• Shoot

Usage

py7dtd_auto_shooting --delay 200

Note: Press ESC to interrupt the bot.

Command line arguments

The following table is listing all the arguments that can be specified:

arg	description	default
help	Arguments description	N/A
delay	Time in ms between each screenshot	500

---

Passcode cracking

Brute-force/dictionary attack on chests/doors passcode.

Usage

It is recommended to set the game in window mode with a resolution of 640x480.

Example of a brute-force attack testing pass codes composed of digits and lowercase characters.

Note: Press ESC to interrupt the bot.

py7dtd_crack_passcode --brute --digits --lower

Get the arguments list with the help function:

py7dtd_crack_passcode --help

Example of a dictionary attack with no limit in tries.

py7dtd_crack_passcode --dict

Note: dictionaries can be found at https://github.com/danielmiessler/SecLists/tree/master/Passwords.

Command line arguments

The available methods are brute-force attack (--brute) and dictionary attack (--dict).

The following table is listing all the arguments to use for each method:

arg	description	default	type
help	Arguments description	N/A	N/A
min	Minimum length	1	brute
max	Maximum length	20	brute
digits	Include digits	False	brute
lower	Include lowercase characters	False	brute
upper	Include uppercase characters	False	brute
lowercyrillic	Include lowercase cyrillic characters	False	brute
uppercyrillic	Include uppercase cyrillic characters	False	brute
special	Include special characters	False	brute
allcharacters	Include all characters	False	brute
dictpath	Dictionary file path	./dictionaries/top1000000.txt	dict
resumedict	Line number to resume a dictionary attack	0	dict
limit	Maximum number of tries	∞	brute, dict
timeout	Maximum time in seconds allowed	∞	brute, dict
delay	Delay in ms between each action	35	brute, dict

---

*If no characters set is selected for the brute-force method, the following sets will be automatically selected: [digits, lower]

Blocks detection

Detection of blocks like topsoil and destroyed stone by analyzing the block's pixel color from the game map.

Usage

Export the game map into a png file with the external tool https://github.com/DorHans/7DTD-SMR. Use the exported map as input for the blocks detections. Each pixel will be analyzed and, if an interested block is detected, it will be marked in red in the output file.

py7dtd_blocks_detection --topsoil --destroyed_stone --input .\map.png

If the argument output is not specified, an output file will be stored in the current folder with the same name as the input file and a suffix _output (e.g. map_output.png).

Get the arguments list with the help function:

py7dtd_blocks_detection --help

Example of detection of topsoil and destroyed stone blocks in the desert biome:

Command line arguments

The following table is listing all the arguments:

arg	description	default
help	Arguments description	N/A
input	Input filename	N/A*
topsoil	Topsoil blocks	False**
dirt	Dirt blocks	False**
gravel	Gravel blocks	False**
destroyed_stone	Destroyed stone blocks	False**
output	Output filename	input + '_output' suffix

*Required

**At least one of these is required

Note: topsoil blocks and dirt blocks appear with the same color in the map.