CN108009598A - Floor plan recognition methods based on deep learning - Google Patents
Floor plan recognition methods based on deep learning Download PDFInfo
- Publication number
- CN108009598A CN108009598A CN201711444198.2A CN201711444198A CN108009598A CN 108009598 A CN108009598 A CN 108009598A CN 201711444198 A CN201711444198 A CN 201711444198A CN 108009598 A CN108009598 A CN 108009598A
- Authority
- CN
- China
- Prior art keywords
- layer
- picture
- model
- floor plan
- deep learning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of floor plan recognition methods based on deep learning, comprise the following steps：（1）The collection and pretreatment of training data；（2）Carry out model training；（3）The use of model.It is an advantage of the invention that：Manually mark is not used, saves manpower；Model structure is simple, trains and using all more time saving；A variety of can meet different scenes demand with interface.
Description
Technical field
The present invention relates to a kind of floor plan recognition methods based on deep learning.
Background technology
Floor plan classification in house property field is also using manual type or simple machine supplementary mode, work efficiency
Extremely low, effect is also very different.
The work efficiency of existing way is extremely low, and cost is high, not only time-consuming and laborious based on artificial classification, but also is difficult to protect
Demonstrate,prove quality, it is also difficult to quality is evaluated and tested and is monitored, it is impossible to the processing of mass data is adapted to, it is difficult in face of the big data epoch
Challenge.
The content of the invention
The defects of to overcome the prior art, the present invention provide a kind of floor plan recognition methods based on deep learning, this hair
Bright technical solution is：
Floor plan recognition methods based on deep learning, comprises the following steps：
（1）The collection and pretreatment of training data；
（2）Carry out model training；
（3）The use of model.
The step（1）Specially：
（1-1）The source of houses pictorial information of multiple house property official websites is acquired using web crawlers, with reference to collection the page rule and
It was found that data source feature, is picture indicia label according to the label information of house property official website, specific acquisition stamp methods are as follows：
A, find mark the picture tag fraternal label, discover whether identify the picture attribute label, if then into
Row obtains；
B, the URL of picture is parsed, is detected whether comprising the label that can identify picture/mb-type, if then being obtained；
（1-2）The data of collection are pre-processed：
The picture of collection is scaled to the picture of short side 100px a,;
B, and then at random the picture of 100 pxX 100px is cut into as training sample;
C, using data enhancement methods, angle rotation is carried out to picture, it is inverse respectively centered on center picture point to each picture
Hour hands are rotated by 90 °, 180 degree and 270 degree；
The step（2）Specially：
（2-1）Trained using pytorch deep learnings frame, the CNN convolutional neural networks that the method for model uses, network is divided into
8 layers, respectively first layer convolutional layer, second layer convolutional layer, third layer pond layer, the 4th layer of convolutional layer, layer 5 convolutional layer, the
Six layers of pond layer, the full context layer of layer 7, the 8th layer of full context layer, layer 7 and the 8th layer of full context layer and label lead to
Cross cross entropy loss function counting loss and carry out backpropagation, by rmsprop optimization method regularized learning algorithm rates, adjusting parameter,
30 epoch are trained altogether.
The step（3）Specially：Api interfaces are provided based on flask frames, in advance preloading step（2）Model,
Comprise the following steps that：
（4-1）The loading function of model is added at flask document entries,
（4-2）Read step（2）In trained model parameter,
（4-3）Build same step（2）In trained model parameter network, and read model parameter initialization net, so
With regard to having reappeared step（2）Trained model,
（4-4）Preserve this model and allow model memory-resident in memory.
The step（3）Specially：Offline batch processing is carried out, is comprised the following steps that：
（5-1）Quick-downloading picture interface is provided, downloads a large amount of pictures in batches,
（5-2）Model provides batch and uses interface in itself, and batch data of reading enter model every time and then batch uses calculation
Method processing, speed is 2~3/second.
It is an advantage of the invention that：Manually mark is not used, saves manpower；Model structure is simple, trains and uses and all compares
It is time saving；A variety of can meet different scenes demand with interface.
Embodiment
The invention will now be further described with reference to specific embodiments, the advantages and features of the present invention will be with description and
It is apparent.But these embodiments are only exemplary, do not form any restrictions to the scope of the present invention.People in the art
Member it should be understood that without departing from the spirit and scope of the invention can to the details of technical solution of the present invention and form into
Row modifications or substitutions, but these modifications and replacement are each fallen within protection scope of the present invention.
The present invention relates to a kind of floor plan recognition methods based on deep learning, comprise the following steps：
（1）The collection and pretreatment of training data；
（2）Carry out model training；
（3）The use of model.
The step（1）Specially：
（1-1）The source of houses pictorial information of multiple house property official websites is acquired using web crawlers, with reference to collection the page rule and
It was found that data source feature, is picture indicia label according to the label information of house property official website, specific acquisition stamp methods are as follows：
A, find mark the picture tag fraternal label, discover whether identify the picture attribute label, if then into
Row obtains；
B, the URL of picture is parsed, is detected whether comprising the label that can identify picture/mb-type, if then being obtained；
（1-2）The data of collection are pre-processed：
The picture of collection is scaled to the picture of short side 100px a,;
B, and then at random the picture of 100 pxX 100px is cut into as training sample;
C, using data enhancement methods, angle rotation is carried out to picture, it is inverse respectively centered on center picture point to each picture
Hour hands are rotated by 90 °, 180 degree and 270 degree；
The step（2）Specially：
（2-1）Trained using pytorch deep learnings frame, the CNN convolutional neural networks that the method for model uses, network is divided into
8 layers, respectively first layer convolutional layer, second layer convolutional layer, third layer pond layer, the 4th layer of convolutional layer, layer 5 convolutional layer, the
Six layers of pond layer, the full context layer of layer 7, the 8th layer of full context layer, layer 7 and the 8th layer of full context layer and label lead to
Cross cross entropy loss function counting loss and carry out backpropagation, by rmsprop optimization method regularized learning algorithm rates, adjusting parameter,
30 epoch are trained altogether.
The step（3）Specially：Api interfaces are provided based on flask frames, in advance preloading step（2）Model,
Comprise the following steps that：
（4-1）The loading function of model is added at flask document entries,
（4-2）Read step（2）In trained model parameter,
（4-3）Build same step（2）In trained model parameter network, and read model parameter initialization net, so
With regard to having reappeared step（2）Trained model,
（4-4）Preserve this model and allow model memory-resident in memory.
The step（3）Specially：Offline batch processing is carried out, is comprised the following steps that：
（5-1）Quick-downloading picture interface is provided, downloads a large amount of pictures in batches,
（5-2）Model provides batch and uses interface in itself, and batch data of reading enter model every time and then batch uses calculation
Method processing, speed is 2~3/second.
Claims (5)
1. the floor plan recognition methods based on deep learning, it is characterised in that comprise the following steps：
（1）The collection and pretreatment of training data；
（2）Carry out model training；
（3）The use of model.
2. the floor plan recognition methods according to claim 1 based on deep learning, it is characterised in that the step
（1）Specially：
（1-1）The source of houses pictorial information of multiple house property official websites is acquired using web crawlers, with reference to collection the page rule and
It was found that data source feature, is picture indicia label according to the label information of house property official website, specific acquisition stamp methods are as follows：
A, find mark the picture tag fraternal label, discover whether identify the picture attribute label, if then into
Row obtains；
B, the URL of picture is parsed, is detected whether comprising the label that can identify picture/mb-type, if then being obtained；
（1-2）The data of collection are pre-processed：
The picture of collection is scaled to the picture of short side 100px a,;
B, and then at random the picture of 100 pxX 100px is cut into as training sample;
C, using data enhancement methods, angle rotation is carried out to picture, it is inverse respectively centered on center picture point to each picture
Hour hands are rotated by 90 °, 180 degree and 270 degree.
3. the floor plan recognition methods according to claim 1 based on deep learning, it is characterised in that the step
（2）Specially：
（2-1）Trained using pytorch deep learnings frame, the CNN convolutional neural networks that the method for model uses, network is divided into
8 layers, respectively first layer convolutional layer, second layer convolutional layer, third layer pond layer, the 4th layer of convolutional layer, layer 5 convolutional layer, the
Six layers of pond layer, the full context layer of layer 7, the 8th layer of full context layer, layer 7 and the 8th layer of full context layer and label lead to
Cross cross entropy loss function counting loss and carry out backpropagation, by rmsprop optimization method regularized learning algorithm rates, adjusting parameter,
30 epoch are trained altogether.
4. the floor plan recognition methods according to claim 1 based on deep learning, it is characterised in that the step
（3）Specially：Api interfaces are provided based on flask frames, in advance preloading step（2）Model, comprise the following steps that：
（4-1）The loading function of model is added at flask document entries；
（4-2）Read step（2）In trained model parameter,
（4-3）Build same step（2）In trained model parameter network, and read model parameter initialization net, so
With regard to having reappeared step（2）Trained model,
（4-4）Preserve this model and allow model memory-resident in memory.
5. the floor plan recognition methods according to claim 1 based on deep learning, it is characterised in that the step
（3）Specially：Offline batch processing is carried out, is comprised the following steps that：
（5-1）Quick-downloading picture interface is provided, downloads a large amount of pictures in batches,
（5-2）Model provides batch and uses interface in itself, and batch data of reading enter model every time and then batch uses calculation
Method processing, speed is 2~3/second.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711444198.2A CN108009598A (en) | 2017-12-27 | 2017-12-27 | Floor plan recognition methods based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711444198.2A CN108009598A (en) | 2017-12-27 | 2017-12-27 | Floor plan recognition methods based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108009598A true CN108009598A (en) | 2018-05-08 |
Family
ID=62061683
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711444198.2A Pending CN108009598A (en) | 2017-12-27 | 2017-12-27 | Floor plan recognition methods based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108009598A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109189963A (en) * | 2018-08-31 | 2019-01-11 | 北京诸葛找房信息技术有限公司 | A kind of source of houses De-weight method based on information of real estate similarity and picture recognition |
| CN109523558A (en) * | 2018-10-16 | 2019-03-26 | 清华大学 | A kind of portrait dividing method and system |
| CN110059750A (en) * | 2019-04-17 | 2019-07-26 | 广东三维家信息科技有限公司 | House type shape recognition process, device and equipment |
| CN110197225A (en) * | 2019-05-28 | 2019-09-03 | 广东三维家信息科技有限公司 | House type spatial match method and system based on deep learning |
| CN110390304A (en) * | 2019-07-24 | 2019-10-29 | 广东南方数码科技股份有限公司 | Automatic classification method, device, electronic equipment and storage medium |
| CN110929319A (en) * | 2019-11-01 | 2020-03-27 | 深圳市彬讯科技有限公司 | Decoration design case information processing method and system |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102054028A (en) * | 2010-12-10 | 2011-05-11 | 黄斌 | Web crawler system with page-rendering function and implementation method thereof |
| CN105574550A (en) * | 2016-02-02 | 2016-05-11 | 北京格灵深瞳信息技术有限公司 | Vehicle identification method and device |
| CN105589943A (en) * | 2015-12-15 | 2016-05-18 | 广州神马移动信息科技有限公司 | Method and device for picture adaptability processing of search result page and server |
| CN105654066A (en) * | 2016-02-02 | 2016-06-08 | 北京格灵深瞳信息技术有限公司 | Vehicle identification method and device |
| CN106530290A (en) * | 2016-10-27 | 2017-03-22 | 朱育盼 | Medical image analysis method and device |
| CN106570515A (en) * | 2016-05-26 | 2017-04-19 | 北京羽医甘蓝信息技术有限公司 | Method and system for treating medical images |
| CN106844614A (en) * | 2017-01-18 | 2017-06-13 | 天津中科智能识别产业技术研究院有限公司 | A kind of floor plan functional area system for rapidly identifying |
| CN107301425A (en) * | 2017-06-09 | 2017-10-27 | 浙江工业大学 | A kind of allowing child daubing methods of marking based on deep learning |
| US20170330068A1 (en) * | 2016-05-16 | 2017-11-16 | Canon Kabushiki Kaisha | Devices, systems, and methods for feature encoding |
-
2017
- 2017-12-27 CN CN201711444198.2A patent/CN108009598A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102054028A (en) * | 2010-12-10 | 2011-05-11 | 黄斌 | Web crawler system with page-rendering function and implementation method thereof |
| CN105589943A (en) * | 2015-12-15 | 2016-05-18 | 广州神马移动信息科技有限公司 | Method and device for picture adaptability processing of search result page and server |
| CN105574550A (en) * | 2016-02-02 | 2016-05-11 | 北京格灵深瞳信息技术有限公司 | Vehicle identification method and device |
| CN105654066A (en) * | 2016-02-02 | 2016-06-08 | 北京格灵深瞳信息技术有限公司 | Vehicle identification method and device |
| US20170330068A1 (en) * | 2016-05-16 | 2017-11-16 | Canon Kabushiki Kaisha | Devices, systems, and methods for feature encoding |
| CN106570515A (en) * | 2016-05-26 | 2017-04-19 | 北京羽医甘蓝信息技术有限公司 | Method and system for treating medical images |
| CN106530290A (en) * | 2016-10-27 | 2017-03-22 | 朱育盼 | Medical image analysis method and device |
| CN106844614A (en) * | 2017-01-18 | 2017-06-13 | 天津中科智能识别产业技术研究院有限公司 | A kind of floor plan functional area system for rapidly identifying |
| CN107301425A (en) * | 2017-06-09 | 2017-10-27 | 浙江工业大学 | A kind of allowing child daubing methods of marking based on deep learning |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109189963A (en) * | 2018-08-31 | 2019-01-11 | 北京诸葛找房信息技术有限公司 | A kind of source of houses De-weight method based on information of real estate similarity and picture recognition |
| CN109523558A (en) * | 2018-10-16 | 2019-03-26 | 清华大学 | A kind of portrait dividing method and system |
| CN110059750A (en) * | 2019-04-17 | 2019-07-26 | 广东三维家信息科技有限公司 | House type shape recognition process, device and equipment |
| CN110197225A (en) * | 2019-05-28 | 2019-09-03 | 广东三维家信息科技有限公司 | House type spatial match method and system based on deep learning |
| CN110390304A (en) * | 2019-07-24 | 2019-10-29 | 广东南方数码科技股份有限公司 | Automatic classification method, device, electronic equipment and storage medium |
| CN110929319A (en) * | 2019-11-01 | 2020-03-27 | 深圳市彬讯科技有限公司 | Decoration design case information processing method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108009598A (en) | Floor plan recognition methods based on deep learning | |
| CN108304765B (en) | Multi-task detection device for face key point positioning and semantic segmentation | |
| Yang et al. | Development of image recognition software based on artificial intelligence algorithm for the efficient sorting of apple fruit | |
| CN105825484B (en) | A kind of depth image denoising and Enhancement Method based on deep learning | |
| CN103578119B (en) | Target detection method in Codebook dynamic scene based on superpixels | |
| CN103226835B (en) | Based on method for tracking target and the system of online initialization gradient enhancement regression tree | |
| CN108875602A (en) | Monitor the face identification method based on deep learning under environment | |
| CN105825191A (en) | Face multi-attribute information-based gender recognition method and system and shooting terminal | |
| CN104408190A (en) | Spark based data processing method and device | |
| CN107392214B (en) | Target detection method based on full-volume integral crack network | |
| CN109410190B (en) | Tower pole reverse-breaking detection model training method based on high-resolution remote sensing satellite image | |
| CN110472652B (en) | Small sample classification method based on semantic guidance | |
| Zhang et al. | Easy domain adaptation method for filling the species gap in deep learning-based fruit detection | |
| CN103258217A (en) | Pedestrian detection method based on incremental learning | |
| CN105468596A (en) | Image retrieval method and device | |
| CN106504207A (en) | A kind of image processing method | |
| CN110287831A (en) | A kind of acquisition methods, device and the electronic equipment at the control point based on terrestrial reference | |
| CN108460418A (en) | A kind of invoice sorting technique based on Text region and semantic analysis | |
| CN107330387A (en) | Pedestrian detection method based on view data | |
| CN104504406A (en) | Rapid and high-efficiency near-duplicate image matching method | |
| CN104376312A (en) | Face recognition method based on word bag compressed sensing feature extraction | |
| CN108717436A (en) | A kind of commodity target method for quickly retrieving based on conspicuousness detection | |
| CN104331717A (en) | Feature dictionary structure and visual feature coding integrating image classifying method | |
| CN112200218A (en) | Model training method and device and electronic equipment | |
| CN109886186A (en) | A kind of face identification method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB03 | Change of inventor or designer information |
Inventor after: Li Zuochao Inventor after: Bai Junfeng Inventor after: Zhang Wenzhan Inventor after: Liu Ziyao Inventor after: Su Weijie Inventor before: Bai Junfeng Inventor before: Zhang Wenzhan Inventor before: Liu Ziyao Inventor before: Su Weijie |
|
| CB03 | Change of inventor or designer information | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180508 |
|
| RJ01 | Rejection of invention patent application after publication |