--- title: What Is mAP (Mean Average Precision)? | daily.dev description: mAP (Mean Average Precision) is the standard evaluation metric for object detection models, combining Precision, Recall, and IoU into a single score. Average... canonical: https://daily.dev/posts/what-is-map-mean-average-precision--0byrlk0yy twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: What Is mAP (Mean Average Precision)? | daily.dev og:description: mAP (Mean Average Precision) is the standard evaluation metric for object detection models, combining Precision, Recall, and IoU into a single score. Average... og:url: https://daily.dev/posts/what-is-map-mean-average-precision--0byrlk0yy og:image: https://api.daily.dev/og/posts/0ByRlk0yy.png og:image:alt: What Is mAP (Mean Average Precision)? 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. # What Is mAP (Mean Average Precision)? **[Medium](https://daily.dev/sources/medium_js)** · 4 min read · 0 upvotes · 0 comments ## Summary mAP (Mean Average Precision) is the standard evaluation metric for object detection models, combining Precision, Recall, and IoU into a single score. Average Precision (AP) is first computed per class, then averaged across all classes to produce mAP. Two common variants are mAP50 (uses a single IoU threshold of 0.50, more lenient) and mAP50-95 (averages across IoU thresholds from 0.50 to 0.95 in 0.05 steps, more rigorous). Scores range from 0 to 1, with values above 0.80 considered good. Improvement strategies include better data quality, accurate annotations, dataset diversity, proper hyperparameter tuning, stronger model architectures, and reducing overfitting. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is the difference between mAP50 and mAP50-95 in object detection? mAP50 evaluates predictions using a single IoU threshold of 0.50, meaning a predicted bounding box only needs to overlap the ground truth by at least 50% to count as correct, making it a more forgiving metric. mAP50-95 instead averages performance across multiple IoU thresholds from 0.50 to 0.95 in increments of 0.05, giving a stricter and more realistic assessment of localization accuracy. _daily.dev surfaces practical breakdowns like this for developers benchmarking object detection models._ ### How do you calculate mAP from per-class Average Precision scores? mAP is the average of the Average Precision (AP) values across all object classes in a dataset. For example, with three classes scoring Person AP = 0.94, Car AP = 0.91, and Bicycle AP = 0.89, mAP equals (0.94 + 0.91 + 0.89) / 3 = 0.91, or 91% overall performance. _Track clear, worked explanations like this on daily.dev when evaluating detection model performance._ ### What mAP score is considered good for an object detection model? An mAP of 0.95 (95%) indicates excellent performance, 0.80 (80%) is good, 0.50 (50%) is acceptable but improvable, and anything below 0.50 is generally considered insufficient. Improving mAP typically involves building a high-quality balanced dataset, creating accurate annotations, increasing dataset diversity, tuning hyperparameters, and using a more capable model architecture. _Developers benchmarking their own detection models can find grounded reference points like this on daily.dev._ ## Similar posts on daily.dev - [What Is Precision?](https://daily.dev/posts/what-is-precision--js5ycsds0) · Medium · 0 upvotes · 0 comments - [How to Evaluate Retrieval Quality in RAG Pipelines \(part 2\): Mean Reciprocal Rank \(MRR\) and Average Precision \(AP\)](https://daily.dev/posts/how-to-evaluate-retrieval-quality-in-rag-pipelines-part-2-mean-reciprocal-rank-mrr-and-average--3ahvydpgt) · Towards Data Science · 0 upvotes · 0 comments - [Machine Learning Model Evaluation: How Do We Know Our Model Is Good?](https://daily.dev/posts/machine-learning-model-evaluation-how-do-we-know-our-model-is-good--qekvejzgn) · Medium · 0 upvotes · 0 comments - [Beyond Accuracy: Why Precision and Recall Matter More Than You Think](https://daily.dev/posts/beyond-accuracy-why-precision-and-recall-matter-more-than-you-think-rnacoqd5w) · BigData Boutique blog · 3 upvotes · 0 comments - [A practical guide to search relevance metrics and evaluation](https://daily.dev/posts/a-practical-guide-to-search-relevance-metrics-and-evaluation-yjcuepxxq) · Meilisearch · 0 upvotes · 0 comments --- Tags: [#machine-learning](https://daily.dev/tags/machine-learning), [#data-science](https://daily.dev/tags/data-science), [#computer-vision](https://daily.dev/tags/computer-vision), [#object-detection](https://daily.dev/tags/object-detection) [View this post on daily.dev](https://daily.dev/posts/what-is-map-mean-average-precision--0byrlk0yy) ```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":"What Is mAP (Mean Average Precision)?","url":"https://daily.dev/posts/what-is-map-mean-average-precision--0byrlk0yy","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/what-is-map-mean-average-precision--0byrlk0yy"},"datePublished":"2026-07-22T10:20:21.390Z","dateModified":"2026-09-14T06:14:19.698Z","description":"mAP (Mean Average Precision) is the standard evaluation metric for object detection models, combining Precision, Recall, and IoU into a single score. Average...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/62b96b0e1573a3435548aea4c955b705?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/62b96b0e1573a3435548aea4c955b705?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Medium","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":"Medium","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/medium","url":"https://daily.dev/sources/medium_js"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/what-is-map-mean-average-precision--0byrlk0yy","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"machine-learning,data-science,computer-vision,object-detection","timeRequired":"PT4M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Medium","item":"https://daily.dev/sources/medium_js"},{"@type":"ListItem","position":3,"name":"What Is mAP (Mean Average Precision)?"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/what-is-map-mean-average-precision--0byrlk0yy#faq","mainEntity":[{"@type":"Question","name":"What is the difference between mAP50 and mAP50-95 in object detection?","acceptedAnswer":{"@type":"Answer","text":"mAP50 evaluates predictions using a single IoU threshold of 0.50, meaning a predicted bounding box only needs to overlap the ground truth by at least 50% to count as correct, making it a more forgiving metric. mAP50-95 instead averages performance across multiple IoU thresholds from 0.50 to 0.95 in increments of 0.05, giving a stricter and more realistic assessment of localization accuracy. daily.dev surfaces practical breakdowns like this for developers benchmarking object detection models."}},{"@type":"Question","name":"How do you calculate mAP from per-class Average Precision scores?","acceptedAnswer":{"@type":"Answer","text":"mAP is the average of the Average Precision (AP) values across all object classes in a dataset. For example, with three classes scoring Person AP = 0.94, Car AP = 0.91, and Bicycle AP = 0.89, mAP equals (0.94 + 0.91 + 0.89) / 3 = 0.91, or 91% overall performance. Track clear, worked explanations like this on daily.dev when evaluating detection model performance."}},{"@type":"Question","name":"What mAP score is considered good for an object detection model?","acceptedAnswer":{"@type":"Answer","text":"An mAP of 0.95 (95%) indicates excellent performance, 0.80 (80%) is good, 0.50 (50%) is acceptable but improvable, and anything below 0.50 is generally considered insufficient. Improving mAP typically involves building a high-quality balanced dataset, creating accurate annotations, increasing dataset diversity, tuning hyperparameters, and using a more capable model architecture. Developers benchmarking their own detection models can find grounded reference points like this on daily.dev."}}]} ```