IVS Application Example: Improving Training Data for Neural Networks
The Intempora Validation Suite (IVS) lets you centralize, store, and share recorded driving scenarios, including the sensor data. You can then index the scenarios, search for particular ones, preview, and postprocess large volumes of driving data on parallel-computing clusters. If you are developing a neural network for object detection or scene understanding, you will need a wealth of training data. With IVS, you can easily preview and filter the data by means of custom tags – even before the data is annotated. With the auto-tagging feature of IVS, newly uploaded data is automatically processed and indexed on the basis of predefined parameters, letting you perform a preliminary text-based search in unlabeled data.
Intelligent Data Filtering
Finding the relevant training data quickly with IVS
How can you benefit from using the IVS? Here is an example: Assume you already have a neural network but it does not perform well enough in certain situations, for example, it struggles to detect trucks underneath gantries. In this case, you can filter the data in the IVS to only show images that contain both trucks and gantries to obtain additional training data for the detector. You can also search for images of drives at night or in the rain.
The IVS is a cloud-based validation tool chain for testing, training, and validating advanced software functions, including perception and deep learning algorithms, against petabytes of recorded data in big-data architectures.
혁신을 추진하세요. 항상 기술 개발의 동향을 주시해야 합니다.
저희 전문 지식 서비스에 가입하세요. dSPACE의 성공적인 프로젝트 사례를 확인해 보세요. 시뮬레이션 및 검증에 대한 최신 정보를 받아보세요. 지금 바로 dSPACE 다이렉트(뉴스레터)를 구독하세요.