AssetStore概要 优惠信息 beta

- 实时检查资源价格变化 -

EasyComputerVision integrates Sentis with a code-free interface to offer an easy way to run computer vision models inside Unity. Currently providing easy image classification and object detection.Implementing machine learning models inside a game engine can be difficult. Unity has helped significantly by creating Sentis, allowing models to be integrated into games. But we go even further by combining it with a no-code interface with EasyComputerVision. No code or knowledge of machine learning is required to be able to run models. The system is fully offilne, operating on a choice of the CPU or GPU, without any reliance on costly cloud services. Simply import your model, use your webcam or an in-game camera as input, and you're ready to make predictions. We offer easy to follow tutorials and have speedy customer service for any queries you may have.EasyComputerVision has been designed for image classification tasks and, in our recent update, object detection. In future updates, we hope to provide tested compatibility for more computer vision types, like image segmentation and depth estimation, so you can run a range of computer vision models. We have recently refactored our code to use the new Unity Sentis, which will allow us to more easily create no-code interfaces that are compatible with more model types.You can view our documentation, roadmap, tutorials, free models, and even make feature requests (we encourage you to!) on our EasyComputerVision GitHub Page.EasyComputerVision is compatible with models in the .onnx format. There are thousands of freely available models to choose from, so we do not provide them within this package. A large database of models can be found at HuggingFace and numerous models are available already in the .onnx format in the newly updated ONNX Model Zoo. EasyComputerVision is currently only compatible with image classification models and, more recently, some object detection models. Please see our GitHub for a list of tested models.If you have any questions, including those regarding model compatibility, please contact us via your preffered method shown on our publisher page.Image classificationObject detectionUse in-game camera as inputUse webcam as inputAutomatically creates a render texture using the model's input dimensionsInference on updateInference on intervalInference on interval using scheduled executionInference on key pressInference on key press using scheduled executionUse a range of engine backend types compatible with GPU and CPUDisplay the input texture as image on a canvasView the output of object detection models in-gameModify the bounding box and text color how you wantView a summary of the model for convenienceScheduled execution can be fine-tuned. The number of model layers processed per frame can be configured, enabling greated control of the inference overhead