CSGO_Aimbot

Overview to train for images

• Place images in object/data foler.
• Place default weights in main directory
• Modify data/object.data and data/object.names file.
• Modify cfg/yolo.config file.
• Run generate.py program that create data.train.txt file.
• Run darknet.exe to start training.

Directory and Files definition

• data/object.names - Contains number of classes,location of train.txt,valid.txt file and location of folder to store training_weights(create the training_weights folder manually in darknet directory).
• data/object.names - Contains the names of the classes to train on. One class per line.
• data/object - Contains images and label to train on.
• data/train.txt - Contains locations of all file to be trained on. i.e. all file in data/object folder

All the above folder and files are to be placed in your darknet/data folder.

Configuration file : cfg/yolov3_CSGO.cfg

Modification to the config fle.

• Comment the batch and subdivision under testing.
• batch - 64 (decrease or increase according to how much your pc can handle.)
• subdivision - 64 (decrease or increase according to how much your pc can handle. Subdivision <= Batch.
• classes - 2 (Your number of classes)
• max_batches - 4000 (2000*classes)
• filter - 21 (classes+5)*3 i.e (2+5)*3
• random - 0/1 1- if you have different size images,0-If all images are of same size

Train yolo model on cs go game images to detect and shoot enemies

• The default weights (darknet53.conv.74) file can be downloaded and placed in the weights folder.

• Modify the generates_train.py file to accept type of images you want to train on(png, jpg) which creates the data/train.txt file

• You can create a validation file (valid.txt) to if you want but we can always test it manually after training.

• Training can be done using the command.

  ./darknet.exe detector train data/object.data cfg/yolov3_custom.cfg weights/darknet53.conv.74*

The next part is converting the model to tensorflow and detect objects on images and videos.