# JunMa11/SegLossOdyssey

A collection of loss functions for medical image segmentation

Repository: https://github.com/JunMa11/SegLossOdyssey
Canonical: https://ross.abutalabs.com/products/seglossodyssey
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2023-11-01T23:10:13+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2652, "days_push": 1036, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4007, forks 610 (observed 2026-08-28T04:08:32.205036+00:00)

## What it is
A curated collection of loss functions for medical image segmentation, accompanying the 'Loss Odyssey in Medical Image Segmentation' survey paper. It catalogs and links implementations of segmentation losses, highlighting compound losses for imbalanced tasks.

## Use cases
- find loss functions for medical image segmentation
- compare segmentation loss functions for imbalanced datasets
- choose a loss for training a U-Net on medical scans
- implement Dice or focal loss variants in PyTorch
- survey recent research on segmentation losses
- improve segmentation performance on small-structure classes

## When to choose
- you are training deep learning models for medical image segmentation
- you need to benchmark or select among many segmentation losses
- your segmentation dataset is highly class-imbalanced

## When to avoid
- you need a maintained training framework rather than a loss catalog
- your task is general object detection or classification
- you need non-PyTorch implementations out of the box

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning
- domain: deep-learning, computer-vision, healthcare, image-processing
- platform: python
- tags: loss-functions, medical-imaging, segmentation, pytorch, research-collection

## Member repositories
- JunMa11/SegLossOdyssey (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:32.205036+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:24:03.143874+00:00, confidence not recorded.
  - readme: https://github.com/JunMa11/SegLossOdyssey (fetched 2026-08-28T04:08:32.205036+00:00, sha cfcaf8263b62)
- Data as of 2026-08-30T08:39:29.467469+00:00.