FIDNet: LiDAR Point Cloud Semantic Segmentation with Fully Interpolation Decoding

Zhao, Yiming; Bai, Lin; Huang, Xinming

doi:10.1109/iros51168.2021.9636385

Cited by 37 publications

(65 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The voxelbased methods [13,14,15] partition the space into a finite number of regions to accelerate the feature extraction process, but suffer serious information loss when resolution is reduced. The projectionbased methods [6,16,17,18,19] also have information loss, but can benefit from some efficient 2D CNN architectures. Due to their efficiency, projection-based methods are more attractive for practical applications.…”

Section: Lidar Segmentationmentioning

confidence: 99%

“…The range value r = x 2 + y 2 + z 2 is calculated according to the point coordinates [x, y, z] T and (u, v) T are image coordinates in the range view. Following the previous works [6,18], the input to our network is a (H, W, 5) tensor R in with channels (x, y, z, rem, r), where rem is the remission or intensity value.…”

Section: Maskrangementioning

confidence: 99%

“…The backbone is to extract low-resolution features from the input R in . Without loss of generality, we use a modified ResNet-34 [30,31,18] as our backbone to extract global features from a range image (see supplementary material Section A for more details). Note that we do not utilize any pre-trained parameters for two reasons.…”

Section: Architecturementioning

confidence: 99%

“…Thus, it would take relatively long time to train the network. To speed up the feed-forward time and the convergence rate, we utilize the Fully Interpolation Decoding (FID) [18] as our decoder module. The FID module is a parameter-free decoder, which only contains bilinear upsampling operation and can facilitate the optimization of our network.…”

Section: Architecturementioning

confidence: 99%

“…To prevent model overfitting, the simple data augmentation schemes are widely used in common practice, including random rotation, translation, flipping, and point dropping [17,18,19]. We also train our model directly with these common data augmentation schemes.…”

Section: Weighted Paste Drop Data Augmentationmentioning

confidence: 99%

See 4 more Smart Citations

MaskRange: A Mask-classification Model for Range-view based LiDAR Segmentation

Gu¹,

Huang²,

Xu³

et al. 2022

Preprint

View full text Add to dashboard Cite

Range-view based LiDAR segmentation methods are attractive for practical applications due to their direct inheritance from efficient 2D CNN architectures. In literature, most range-view based methods follow the per-pixel classification paradigm. Recently, in the image segmentation domain, another paradigm formulates segmentation as a mask-classification problem and has achieved remarkable performance. This raises an interesting question: can the mask-classification paradigm benefit the range-view based LiDAR segmentation and achieve better performance than the counterpart per-pixel paradigm? To answer this question, we propose a unified mask-classification model, MaskRange, for the rangeview based LiDAR semantic and panoptic segmentation. Along with the new paradigm, we also propose a novel data augmentation method to deal with overfitting, context-reliance, and class-imbalance problems. Extensive experiments are conducted on the SemanticKITTI benchmark. Among all published range-view based methods, our MaskRange achieves state-of-the-art performance with 66.10 mIoU on semantic segmentation and promising results with 53.10 PQ on panoptic segmentation with high efficiency. Our code will be released.

show abstract