GMFE-Net: point cloud semantic segmentation with general multi-feature fusion and extraction
Bibliographic record
Abstract
Semantic segmentation serves an essential role in the comprehension of 3D scenes. Existing methods mostly adopt a U-Net architecture, where point cloud feature maps are extracted during the encoder stage and fused multi-scale features using transposed convolution in the decoder stage. However, up-sampling with transposed convolution loses some detailed features. To mitigate this issue and enhance feature learning, we propose the GMFE-Unit, a novel module designed for comprehensive feature learning. The GMFE-Unit comprises three synergistic components: (1) a Local Multi-feature Fusion and Extraction (LocMFE) block that concurrently fuses fine-grained geometric features (e.g. coordinates, distances, angles) with diverse attribute features (e.g. RGB, intensity); (2) an Attention-pooling block that adaptively weights these fused features to prioritize the most critical information for segmentation; and (3) a Global Multi-feature Fusion and Extraction (GloMFE) block that captures high-level scene context by computing volume ratios and distance variations between local neighbourhoods and the global scene. Based on this powerful unit, we develop GMFE-Net, which applies GMFE-Unit in both encoder and decoder stages. Notably, this architectural choice transforms the decoder from a passive up-sampling tool into an active feature refinement and fusion engine, replacing traditional transposed convolution to significantly mitigate information loss. We evaluated GMFE-Net and achieved high semantic segmentation accuracy across three datasets from different sources, SensatUrban, DALES, and EVO. Notably, our method achieves significant gains in underrepresented categories. Furthermore, we conduct the first comprehensive analysis of how various point cloud attribute features (RGB, intensity, echo deviation, etc.) impact semantic segmentation, revealing the underlying reasons for their influence on performance. The code of our key contributions is publicly accessible at: https://github.com/LiTao-0797/GMFE-Net.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".