FREMONT, CA – November 7, 2017 – SKYSITE is pleased and excited to announce that the ARC / SKYSITE Development team has a total of four papers being presented at an upcoming premier conference and workshop focusing on document analysis and recognition.
With several impactful releases this year, including Submittals (March) and the Big Data Back End (July) it’s evident that the team behind SKYSITE constantly strives to make the best Construction and Facilities’ Document Management platform available. However, what is little known is the level of expertise of the team behind the app; they are truly masters at OCR and recognized in the field.
Proof of this is earlier in November, members of the Development Team traveled to Kyoto, Japan to present a total of four papers. One paper was presented at ICDAR2017, the 14th IAPR International Conference on Document Analysis and Recognition. The ICDAR conference is most recognized and prestigious conference in the document analysis domain.
The conference and technical program hosted traditional and new workshops and tutorials on topics aimed at the study of cutting-edge problems in the field. Researchers had the opportunity to learn new algorithms as well as results of numerous competitions that are driving interest in the community.
Attendees and presenters represented prestigious universities (University of California Santa Cruz, University of Innsbruck, Austria, Indian Statistical Institute, Kolkata, India to name a few), and innovative software companies such as Adobe Research, India, Microsoft Research Asia, Nokia Bell Laboratories and more.
The three other papers were given at a workshop, GREC 2017: 12th Annual International Workshop on Graphics Recognition, held in conjunction with ICDAR2017. This annual workshop provides an opportunity for researchers and practitioners to share new ideas and knowledge about graphics recognition methods.
In Japan, the papers were presented by Purnendu Banerjee and Supriya Das. However, huge congratulations go out to the entire team on a phenomenal job in the research, writing and acceptance of the papers: Purnendu Banerjee, Sumit Choudhary, Supriya Das, Himadri Majumder, Srinivas Mukkamala, Rahul Roy, B. B. Chaudhuri, Bhagesh Seraogi, and Sandip Kumar Maity.
It should be noted that this isn’t the first time for the team to present their expertise at an international conference and that previous papers have been published in IEEE.
We would like to congratulate our team on a phenomenal job in the research, writing and acceptance of the papers. A full list of the papers and contributors, along with a synapsis of each paper and links to the shows are below.
The 14th IAPR International Conference on Document Analysis and Recognition Kyoto, Japan (ICDAR2017)
November 13-15, 2017
A System for Creating Automatic Navigation among Architectural and Construction Documents
Purnendu Banerjee, Sumit Choudhary, Supriya Das, Himadri Majumder, Srinivas Mukkamala, Rahul Roy, B. B. Chaudhuri
The topic of this paper, an improved and robust version of OCR system, is introduced in SKYSITE. From uploaded documents, it can now efficiently extract different and complex reference symbols (e. g. textured, noisy, broken, oriented etc.) including circular and pentagonal call-outs. It can even recognize the text within the symbols accurately.
This change will significantly boost the existing auto-hyperlinking mechanism in terms of speed as well as accuracy, requiring less manual intervention.
Nov. 9-10, 2017
Automatic Elevation Datum Detection and Hyperlinking of Architecture, Engineering & Construction Documents
Purnendu Banerjee, Supriya Das, Bhagesh Seraogi, Himadri Majumdar, Srinivas Mukkamala, Rahul Roy
This paper presents how not only call-outs but elevation datum symbols will be automatically distinguished through our enhanced OCR system. In addition, the region around the symbol will be recursively examined to acquire the destination document. Finally, the destination document is hyperlinked with the desired document.
Automatic Orientation Correction of AEC Drawing Documents
Bhagesh Seraogi, Supriya Das, Purnendu Banerjee, Himadri Majumdar, Srinivas Mukkamala, Rahul Roy
This paper examines how, due to the presence of differently oriented documents (i.e. 90°, 180°, 270°), the recognition accuracy of the existing OCR system suffers and the document viewing experience of the user becomes clumsy. Our recent enhancement enables the existing OCR system to automatically identify and correct the orientation of the document, thus improving the user experience.
A Novel Approach for Detecting Circular Callouts in AEC Drawing Documents
Sandip Kumar Maity, Bhagesh Seraogi, Purnendu Banerjee, Supriya Das, Himadri Majumdar, Srinivas Mukkamala, Rahul Roy
The focus of this paper is the time complexity of the conventional Hough Transform based circular symbol detection mechanism is quite high. Therefore, we have developed a novel circular symbol detection mechanism which reduces the time complexity of the existing circular symbol detection approach, enambling the auto-hyperlinking mechanism will perform faster.