Open abhigoku10 opened 1 year ago
The purpose of making the clean models available, without being enmeshed in a training codebase, is to enable and encourage incorporation of these models into existing training pipelines. For training semantic segmentation, for instance, you could plug these models into one of the many available codebases, such as https://github.com/NVIDIA/semantic-segmentation
@mynameistechno @ggolin @TiRune can you guys share the training pipeline ? Thanks in advance