Requirements for completing these tasks: computer, webcam, Google Mail account.
There are three learning strategies to choose from:
- Study each exercise carefully and perform each one.
- Choose one topic and focus on studying it. In the best case, we’ll start writing a joint research article (academic paper) on the chosen topic.
- Combined option. Quickly familiarize yourself with the proposed tasks and concentrate on studying one topic. With the possibility of writing a research article.
Most likely we’ll choose the second strategy (option B) for topics 7 and 8.
The following topics are available for study:
- Classification of three-dimensional (3D) medical images.
File
01_3D_image_classification.ipynb. - Implementation and launch of a Generative Adversarial Network (GAN).
A simple version of GAN and a more advanced one.
Files
02.1_Simple_GAN_example.ipynband02.2_DCGAN_example.ipynb. - Image processing with Python without neural networks.
File
03_Image_processing_with_Python.ipynb. - OpenCV features detectors and descriptor extractors algorithms with GUI.
File
04_OpenCV_feature_detectors_and_descriptor_extractors.ipynb. - Implementation and experiments with ReAct (reasoning and acting) AI Agent.
File
05_AI_Agent.ipynb. - Generating heat maps using neural network class activation maps.
File
06_Heatmap_using_CAM.ipynb. - Independent work on assignment.
Self-supervised contrastive learning code example.
Directory
./additional_data/Contrastive_Learning_research. - Independent work on assignment.
Maps segmentation using the pipeline in file
./additional_data/UNet_image_segmentation_v2.ipynb. Maps generation using the pipeline in file./additional_data/Pix2Pix_GAN_implementation.ipynb.
After your internship, you can continue your studies with the following course: "Practical Deep Learning for Coders" (https://course.fast.ai/).
P.S. There is no models directory, because models took up too much space.
P.P.S. Simpler tasks are placed in simple_tasks directory.