SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2002.00718 · 2020

Modeling the Background for Incremental Learning in Semantic Segmentation

arXiv · PDF · Open in the Atlas

Code that ran

5 of 7 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 2 failures of our instrument; 2 no recorded run), measured 2026-11-21; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2002.00718.json · badge: /api/ran/2002.00718.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
fcdl94/MiB community (classifier label)
classifier basis not recorded on the link
5 ran · 2 no recorded run
FunctionStatusWhere it lives
filter_images Ran
ran on a synthesized input
fcdl94/MiB/dataset/utils.py
code served (permissive licence) · get_code("21c634fc9f57dd4c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

class MockDataset:
    def __init__(self, size=10):
        self.size = size
    
    def __len__(self):
        return self.size
    
    def __getitem__(self, idx):
        # Return (image, mask) tuples
        # mask is a numpy array with class labels
        image = np.random.randint(0, 256, (32, 32, 3), dtype=np.uint8)
        if idx % 3 == 0:
            mask = np.array([[1, 2], [3, 4]], dtype=np.uint8)
        elif idx % 3 == 1:
            mask = np.array([[2, 5], [6, 7]], dtype=np.uint8)
        else:
            mask = np.array([[1, 5], [2, 6]], dtype=np.uint8)
        return (image, mask)

dataset = MockDataset(size=5)
labels = [1, 2, 3]
labels_old = [4, 5]
overlap = True

ARGS = {
    "dataset": dataset,
    "labels": labels,
    "labels_old": labels_old,
    "overlap": overlap
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family int — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
flip Ran
ran; our contract draft was wrong, not the code
fcdl94/MiB/segmentation_module.py
pointer only (licence: MIT); the stored copy get_code serves, from zhengyuan-xie/ECCV24_NeST, is pointer only (licence: NONE) · get_code("7f8043e7dd6c2735")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from zhengyuan-xie/ECCV24_NeST, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_loss Ran
ran; our contract draft was wrong, not the code
fcdl94/MiB/utils/loss.py
code served (permissive licence) · get_code("1d5ab3d2728d692e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned FocalLoss. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Define test arguments for get_loss function
ARGS = {
    'loss_type': 'focal_loss'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type FocalLoss — a FocalLoss output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a FocalLoss output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
group_images Ran
ran on a synthesized input
fcdl94/MiB/dataset/utils.py
code served (permissive licence) · get_code("afc11e8a9f0184d3")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import torch

# Create a minimal dataset
# The function expects dataset[i][1] to be array-like (the label/mask component)
# and calls np.unique() on it, so we need array-like objects

class MinimalDataset:
    def __init__(self, size=3):
        self.size = size
    
    def __len__(self):
        return self.size
    
    def __getitem__(self, idx):
        # Return (image, mask) tuples
        # mask should be array-like with class labels
        if idx == 0:
            mask = np.array([1, 2])  # Contains labels 1 and 2
        elif idx == 1:
            mask = np.array([2, 3])  # Contains labels 2 and 3
        else:
            mask = np.array([0, 255])  # Contains background and ignore index
        
        image = np.zeros((4, 4))  # Dummy image
        return (image, mask)

dataset = MinimalDataset(size=3)
labels = [1, 2, 3]

ARGS = {
    "dataset": dataset,
    "labels": labels
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
modify_command_options Ran
ran on a synthesized input
fcdl94/MiB/argparser.py
code served (permissive licence) · get_code("dcde5d7316c65773")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned OptsConfig. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

class OptsConfig:
    def __init__(self):
        self.dataset = 'voc'
        self.num_classes = None
        self.visualize = True
        self.sample_num = 5
        self.method = 'LWF'
        self.loss_kd = None
        self.loss_de = None
        self.icarl = False
        self.icarl_importance = None
        self.regularizer = None
        self.reg_importance = None
        self.unce = False
        self.unkd = False
        self.init_balanced = False
        self.overlap = True
        self.cross_val = True
        self.no_overlap = None
        self.no_cross_val = None

ARGS = {
    'opts': OptsConfig()
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type OptsConfig — a OptsConfig output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a OptsConfig output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_per_task_classes Not yet run
unverified (harvested, no recorded run)
fcdl94/MiB/tasks.py
code served (permissive licence) · get_code("1803646c9607e58b")
How it ran

no invocation recorded.

get_task_labels Not yet run
unverified (harvested, no recorded run)
fcdl94/MiB/tasks.py
code served (permissive licence) · get_code("ad850a613acafcc0")
How it ran

no invocation recorded.

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 3 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 3 later) and 4 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
flip 4 other papers
4 later
2010.07092 (2020-10)
Data Augmentation for Meta-Learning
2210.07207 (2022-10)
Attribution-aware Weight Transfer: A Warm-Start Initialization for Class-Incremental Semantic Segmentation
aaai_28394 (2024 (year only))
DTMFormer: Dynamic Token Merging for Boosting Transformer-Based Medical Image Segmentation
and 1 more not named here
get_loss 2 other papers
2 later
Cermelli_Incremental_Learning_in_Semantic_Segmentation_From_Image_Labels_CVPR_2022_paper (2022 (year only))
Incremental Learning in Semantic Segmentation from Image Labels
2210.07207 (2022-10)
Attribution-aware Weight Transfer: A Warm-Start Initialization for Class-Incremental Semantic Segmentation
group_images 1 other paper
1 later
2106.00472 (2021-06)
Detecting Anomalies in Semantic Segmentation with Prototypes

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-21.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Despite their effectiveness in a wide range of tasks, deep architectures suffer from some important limitations. In particular, they are vulnerable to catastrophic forgetting, i.e. they perform poorly when they are required to update their model as new classes are available but the original training set is not retained. This paper addresses this problem in the context of semantic segmentation. Current strategies fail on this task because they do not consider a peculiar aspect of semantic segmentation: since each training step provides annotation only for a subset of all possible classes, pixels of the background class (i.e. pixels that do not belong to any other classes) exhibit a semantic distribution shift. In this work we revisit classical incremental learning methods, proposing a new distillation-based framework which explicitly accounts for this shift. Furthermore, we introduce a novel strategy to initialize classifier's parameters, thus preventing biased predictions toward the background class. We demonstrate the effectiveness of our approach with an extensive evaluation on the Pascal-VOC 2012 and ADE20K datasets, significantly outperforming state of the art incremental learning methods.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2002.00718")
get_code_for_paper("2002.00718")
have("2002.00718")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2002.00718.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2002.00718.svg)](https://syntology.ai/paper/2002.00718)

Connect an agent — have() is free.