Pretraining boosts out-of-domain robustness for pose estimation
This work addresses robustness challenges in pose estimation for real-world applications with limited data, though it is incremental as it confirms known benefits of transfer learning in a new context.
The study tackled the problem of out-of-domain robustness in pose estimation, particularly for small training sets, by testing architectures like MobileNetV2s, ResNets, and EfficientNets on a dataset of 30 horses and a new Horse-C benchmark, showing that pretraining on ImageNet improves performance on both within-domain and out-of-domain data, with better ImageNet models generalizing better across animal species.
Neural networks are highly effective tools for pose estimation. However, as in other computer vision tasks, robustness to out-of-domain data remains a challenge, especially for small training sets that are common for real-world applications. Here, we probe the generalization ability with three architecture classes (MobileNetV2s, ResNets, and EfficientNets) for pose estimation. We developed a dataset of 30 horses that allowed for both "within-domain" and "out-of-domain" (unseen horse) benchmarking - this is a crucial test for robustness that current human pose estimation benchmarks do not directly address. We show that better ImageNet-performing architectures perform better on both within- and out-of-domain data if they are first pretrained on ImageNet. We additionally show that better ImageNet models generalize better across animal species. Furthermore, we introduce Horse-C, a new benchmark for common corruptions for pose estimation, and confirm that pretraining increases performance in this domain shift context as well. Overall, our results demonstrate that transfer learning is beneficial for out-of-domain robustness.