MergeNet: A Deep Net Architecture for Small Obstacle Discovery
This addresses the challenge of scaling data annotation for small obstacles in autonomous driving, offering a more data-efficient solution.
The paper tackles the problem of small obstacle discovery in autonomous driving by proposing MergeNet, a novel network architecture that achieves state-of-the-art results on the Lost and Found dataset using only 135 images, compared to 1000 images in previous benchmarks, and shows comparable performance to methods trained on 6000 images with only 1000 samples.
We present here, a novel network architecture called MergeNet for discovering small obstacles for on-road scenes in the context of autonomous driving. The basis of the architecture rests on the central consideration of training with less amount of data since the physical setup and the annotation process for small obstacles is hard to scale. For making effective use of the limited data, we propose a multi-stage training procedure involving weight-sharing, separate learning of low and high level features from the RGBD input and a refining stage which learns to fuse the obtained complementary features. The model is trained and evaluated on the Lost and Found dataset and is able to achieve state-of-art results with just 135 images in comparison to the 1000 images used by the previous benchmark. Additionally, we also compare our results with recent methods trained on 6000 images and show that our method achieves comparable performance with only 1000 training samples.