TinyMind is a deep learning framework developed using unity that does not rely on any third-party libraries.The custom SRP rendering pipeline has not been tested.Deep learning has become one of the popular development directions in the field of science today, Image recognition, speech recognition, text translation, object detection can all be solved using deep learning.However, to develop deep learning applications, in most cases only Python can be used. Now, this situation will change. With TinyMind, you can develop deep learning applications in Unity using efficient and elegant C # language.With TinyMind, you can perform image recognition, object detection, image generation, semantic segmentation and reinforcement learning, and more.Deep learning is based on big data, so it is inevitable to use third-party datasets, TinyMind uses the following third-party datasets in the demos:Asset uses FashionMNIST, udacity_driving, CeyMo, P1_Facial_Keypoints under MIT license, see Third-Party Notices.txt file in package for details.Dynamic Computation Graphs and Automatic Differentiation. This is the most important feature.Modular design allows you to easily build neural networks.Perform high performance computations using Burst and Job systems.Provides common transforms to improve the accuracy of neural network training and inference.Weight loading and saving.