LightMIS: 0.13M-Param Medical Segmentation Net Cuts 99% Params, Matches Accuracy
Andrei Arhire · hf · 2026-09-28
A new HF project introduces LightMIS, a family of ultra-lightweight CNNs for 2D binary medical image segmentation without a learned stage-wise decoder, using Scale-Aligned Projection blocks and an Adaptive Fusion Cascade.
Evaluated under the nnU-Net v2.3.1 five-fold protocol on DRIVE, Kvasir-SEG, DSB18, BUSI, ISIC-2017/2018:
- Full LightMIS has just 0.131M parameters and 0.575 GFLOPs (256×256 input), scoring 86.71% Dice / 78.99% IoU macro-average — within 0.04/0.08 points of Mobile U-ViT;
- It cuts parameters by 90.58–99.61% and GFLOPs by 82.54–96.14% vs. Mobile U-ViT, nnWNet, and nnU-Net;
- All variants achieve full GPU delegation on an Arm Mali-G52 MC2 with 53.31–138.31 ms median latency, demonstrating on-device feasibility. Code is open source.
More from Research
- Goodfire grants geometric_intel lab funding for AI interpretability research — ninamiolane · 2026-09-29
- Bespoke Labs Launches AutoResearchExam Benchmark, Again, With a Demo Video — gregd_nlp · 2026-09-28
- Agentick benchmark accepted at NeurIPS: LLM vs RL agents on same tasks, no single winner — pcastr · 2026-09-28
- IROS 2026 has 1,933 papers — researcher curates 130-paper reading list on VLA and robot learning — GlenBerseth · 2026-09-28
- Ex-game-AI developer: general agents are taking over bespoke game AI systems — weballergy · 2026-09-28
- Kaggle Game Arena: Google's LLM benchmark pits models against each other in chess, poker, werewolf — weballergy · 2026-09-28