[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84933-en":3,"doc-seo-84933-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},84933,687197207639,"Asher","https://ap-avatar.wpscdn.com/davatar_a8503ba1806abce46bf441b54a3ca4cd",8,"Research & Report","MiLSD: A Micro Line-Segment Detector for Resource-Constrained Devices","Line segment detection underpins visual SLAM, 3D reconstruction, and industrial inspection, yet current deep models often exceed the memory limits of low-cost MCUs. MiLSD studies the best attainable accuracy within a sub-megabyte budget by proposing an MCU-tailored detector and comparing three compact output representations using a fully-convolutional backbone. Experiments show F-Clip center-with-length-and-angle works best, with 8-bit quantization maintaining full-precision while 4-bit degrades angle regression; recovery via QAT is partial. With a one-megabyte activation budget and inference enhancements, MiLSD raises sAP10 on ShanghaiTech Wireframe from 10.6 to 24.1, mapping accuracy–memory trade-offs for embedded vision.","MiLSD: A Micro Line-Segment Detector for Resource-Constrained Devices  \nParsa Hassani Shariat Panahiπ , Amir Hossein Jalilvandπ , and M. Hassan Najafi+   \narXiv :2607 .06600v 1 [ cs .CV] 7 Jul 2026  \nAbstract—Line segment detection is a key building block in visual SLAM, 3D reconstruction, and industrial inspection. Recent deep learning methods have greatly improved accuracy, yet even the smallest models require several megabytes of memory, exceeding low-cost MCU capacity. This work investigates the maximum achievable accuracy under a sub-megabyte budget. We propose MiLSD, a detector tailored for MCU-level constraints, and systematically compare three output representations within a compact fully-convolutional backbone.  \nOur study shows that the proposed F-Clip center-with-lengthand-angle formulation learns most effectively at small model sizes. We find that 8-bit quantization preserves full-precision performance, while 4-bit quantization causes significant degradation, particularly in angle regression, with quantization-aware training recovering only part of the loss. With a one-megabyte activation budget and inference enhancements including subpixel decoding, test-time augmentation, and a lightweight verifier, MiLSD improves sAP10 on ShanghaiTech Wireframe from 10.6 (25k parameters, 0.25 MB) to 24.1 within 1 MB. Rather than competing with GPU-scale parsers, we map the accuracy– memory trade-off across representations, bit-widths, capacities, and post-processing strategies for embedded vision systems.  \nIndex Terms—line segment detection, TinyML, quantization, memory-constrained inference, embedded vision.  \nI. INTRODUCTION  \nLINE segments are a primitive structural feature in com  \nputer vision: the straight edges of walls, doors, buildings, and machined parts [1] . They support SLAM, structure-frommotion, vanishing-point estimation, lane and power-line detection, and industrial inspection. While much recent progress has targeted GPU-or cloud-based platforms, this work focuses on detection under the tight memory and compute constraints of real-time embedded hardware.  \nClassical detectors such as LSD [1](Line Segment Detector) and EDLines [2] run on a CPU but find all edges, while modern learned wireframe parsers [3]–[6] recover only salient segments but require GPU- or phone-class compute. Table I situates representative methods across this spectrum. Classical detectors grow line-support regions from local gradients and validate them statistically. LSD groups pixels with consistent gradient orientation and accepts a segment if its number of false alarms is below one. EDLines reaches comparable quality faster by chaining edge pixels into clean chains. ELSED [7] targets embedded CPUs for high frame rates. Their shared weakness: accuracy degrades under blur, low contrast, and clutter, and runtime is content-dependent. At its evaluated  \nπ School of Computer Engineering, Iran University of Science and Technology, Tehran, Iran  \n+ Electrical, Computer, and Systems Engineering Department, Case Western Reserve University, OH, USA  \n640 × 480 resolution, ELSED also requires several full-frame gradient and edge buffers (∼ 1.5–2 MB) that exceed even the 1 MB SRAM of an STM32H7, and since its edge walk is global and data-dependent, it cannot be tiled and admits no static worst-case memory bound, unlike a fixed-cost CNN. Classical detectors have also been mapped to FPGAs and ASICs for deterministic latency, but these implementations accelerate hand-designed gradient logic, not neural networkson general-purpose microcontrollers.  \nShanghaiTech Wireframe [8] reframed line detection as a learning problem. L-CNN [3] proposed junctions and verified candidate lines. AFM [9] used attraction fields, while HAWP [4] combined holistic fields with endpoint verification. ULSD [10] generalized across pinhole, fisheye, and spherical cameras, and LETR [11] uses transformers for direct line detection. L-CNN, HAWP, ULSD, and LETR achieve sAP 10 ≈ 63–70","cbCaikfce2140a3Y","https://ap.wps.com/l/cbCaikfce2140a3Y","pdf",1643217,2,1,10,"English","en",105,"# Introduction\n## Line segment detection background\n## Limits of classical and learned methods\n## Memory constraints on MCUs\n## Prior encoding and quantization approaches\n# Proposed approach (MiLSD)","[{\"question\":\"What problem does MiLSD address?\",\"answer\":\"MiLSD targets line segment detection with accuracy high enough for applications like SLAM and inspection, while operating under very tight memory and compute constraints typical of microcontrollers.\"},{\"question\":\"Which output representation performs best for small model sizes?\",\"answer\":\"The F-Clip style center-with-length-and-angle formulation learns most effectively when the model is constrained to small sizes.\"},{\"question\":\"How does quantization affect MiLSD performance?\",\"answer\":\"8-bit quantization preserves near full-precision performance, while 4-bit quantization causes substantial degradation, especially in angle regression; quantization-aware training recovers only part of the lost accuracy.\"}]",1784199466,25,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"milsd-a-micro-line-segment-detector-for-resource-constrained-devices","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,47,50],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":20},"https://docshare.wps.com/document/","Document",{"item":48,"name":12,"@type":43,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/milsd-a-micro-line-segment-detector-for-resource-constrained-devices/84933/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-24","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does MiLSD address?","Question",{"text":75,"@type":76},"MiLSD targets line segment detection with accuracy high enough for applications like SLAM and inspection, while operating under very tight memory and compute constraints typical of microcontrollers.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Which output representation performs best for small model sizes?",{"text":80,"@type":76},"The F-Clip style center-with-length-and-angle formulation learns most effectively when the model is constrained to small sizes.",{"name":82,"@type":73,"acceptedAnswer":83},"How does quantization affect MiLSD performance?",{"text":84,"@type":76},"8-bit quantization preserves near full-precision performance, while 4-bit quantization causes substantial degradation, especially in angle regression; quantization-aware training recovers only part of the lost accuracy.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,134],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":22,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":22,"slug":133},"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]