--- title: MLfix to quickly fix datasets | daily.dev description: MLfix is an open-source tool developed by Collabora that combines unsupervised ML pipelines with a novel UI to help annotators and ML engineers identify and... canonical: https://daily.dev/posts/mlfix-to-quickly-fix-datasets-crm8zr7lj twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: MLfix to quickly fix datasets | daily.dev og:description: MLfix is an open-source tool developed by Collabora that combines unsupervised ML pipelines with a novel UI to help annotators and ML engineers identify and... og:url: https://daily.dev/posts/mlfix-to-quickly-fix-datasets-crm8zr7lj og:image: https://api.daily.dev/og/posts/Crm8ZR7Lj.png og:image:alt: MLfix to quickly fix datasets og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # MLfix to quickly fix datasets **[Collabora Blog](https://daily.dev/sources/collabora)** · 6 min read · 0 upvotes · 0 comments ## Summary MLfix is an open-source tool developed by Collabora that combines unsupervised ML pipelines with a novel UI to help annotators and ML engineers identify and remove labelling errors in image-based datasets. The post demonstrates MLfix's effectiveness on synthetic datasets generated via the Carlafox/CARLA simulator for vehicle and traffic sign detection, uncovering issues like incorrect bounding box widths, opposite-facing traffic signs, ambiguous class labels, fully occluded objects, and zero-width cyclist annotations — errors that would have been nearly impossible to catch manually. The tool is available on GitHub. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How can I automatically detect labelling errors in a large ML image dataset without manually reviewing every sample? MLfix combines unsupervised ML pipelines with a visual UI that groups similar samples, making it straightforward to spot systematic errors like incorrect bounding box dimensions, opposite-facing objects, occluded annotations, and ambiguous class labels. It was used to find critical issues in CARLA-generated synthetic datasets that were assumed to be error-free, including zero-width bounding boxes and duplicate-looking labels for the same class. _ML engineers auditing training data find the latest dataset tooling discussions on daily.dev._ ### Are synthetic datasets generated from simulators like CARLA guaranteed to be error-free compared to human-labelled data? No — synthetic datasets from CARLA are not error-free. Running CARLA-generated datasets through MLfix revealed multiple systematic issues: incorrect bounding box widths for partially visible traffic signs, opposite-facing signs being annotated, two visually distinct signs sharing the same class label, fully occluded objects being annotated, and cyclist bounding boxes with zero width due to a suspected CARLA API bug. _Developers building autonomous driving pipelines track simulator and dataset tooling news on daily.dev._ ## Similar posts on daily.dev - [Improving annotation quality with machine learning](https://daily.dev/posts/improving-annotation-quality-with-machine-learning-bhvf4m2b8) · InfoWorld · 0 upvotes · 0 comments - [Finding Bugs using LLMs](https://daily.dev/posts/finding-bugs-using-llms-pbp8oopzr) · Materialize · 0 upvotes · 0 comments - [Beyond ‘Shift Left’: Improving AI Training Data](https://daily.dev/posts/beyond-shift-left-improving-ai-training-data-hebwp7ld5) · The New Stack · 2 upvotes · 0 comments - [Leveraging PyFixest for High-Cardinality Marketplace Modeling at Instacart](https://daily.dev/posts/leveraging-pyfixest-for-high-cardinality-marketplace-modeling-at-instacart-kmte6zxgg) · Instacart · 0 upvotes · 0 comments --- Tags: [#machine-learning](https://daily.dev/tags/machine-learning), [#data-science](https://daily.dev/tags/data-science), [#object-detection](https://daily.dev/tags/object-detection) [View this post on daily.dev](https://daily.dev/posts/mlfix-to-quickly-fix-datasets-crm8zr7lj) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"MLfix to quickly fix datasets","url":"https://daily.dev/posts/mlfix-to-quickly-fix-datasets-crm8zr7lj","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/mlfix-to-quickly-fix-datasets-crm8zr7lj"},"datePublished":"2026-08-11T08:55:17.051Z","dateModified":"2026-08-11T09:27:12.497Z","description":"MLfix is an open-source tool developed by Collabora that combines unsupervised ML pipelines with a novel UI to help annotators and ML engineers identify and...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/2add6e254bba9aaa2286205d25fd85cc?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/2add6e254bba9aaa2286205d25fd85cc?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Collabora Blog","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Collabora Blog","logo":"https://media.daily.dev/image/upload/s--ZGZ0FjqS--/f_auto,q_auto/v1786438306/logos/collabora?_a=BAMAMicg0","url":"https://daily.dev/sources/collabora"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/mlfix-to-quickly-fix-datasets-crm8zr7lj","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"machine-learning,data-science,object-detection","timeRequired":"PT6M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collabora Blog","item":"https://daily.dev/sources/collabora"},{"@type":"ListItem","position":3,"name":"MLfix to quickly fix datasets"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/mlfix-to-quickly-fix-datasets-crm8zr7lj#faq","mainEntity":[{"@type":"Question","name":"How can I automatically detect labelling errors in a large ML image dataset without manually reviewing every sample?","acceptedAnswer":{"@type":"Answer","text":"MLfix combines unsupervised ML pipelines with a visual UI that groups similar samples, making it straightforward to spot systematic errors like incorrect bounding box dimensions, opposite-facing objects, occluded annotations, and ambiguous class labels. It was used to find critical issues in CARLA-generated synthetic datasets that were assumed to be error-free, including zero-width bounding boxes and duplicate-looking labels for the same class. ML engineers auditing training data find the latest dataset tooling discussions on daily.dev."}},{"@type":"Question","name":"Are synthetic datasets generated from simulators like CARLA guaranteed to be error-free compared to human-labelled data?","acceptedAnswer":{"@type":"Answer","text":"No — synthetic datasets from CARLA are not error-free. Running CARLA-generated datasets through MLfix revealed multiple systematic issues: incorrect bounding box widths for partially visible traffic signs, opposite-facing signs being annotated, two visually distinct signs sharing the same class label, fully occluded objects being annotated, and cyclist bounding boxes with zero width due to a suspected CARLA API bug. Developers building autonomous driving pipelines track simulator and dataset tooling news on daily.dev."}}]} ```