| Predicate |
Object |
| assignee |
http://rdf.ncbi.nlm.nih.gov/pubchem/patentassignee/MD5_819d2d00412d671253120d16a2f09908 |
| classificationCPCAdditional |
http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06T2207-20221
http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06T2207-20084
http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06T2207-20081
http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06T2207-10028 |
| classificationCPCInventive |
http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06T3-4038
http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06T5-50
http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06N3-08
http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06T9-002
http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06T17-00 |
| classificationIPCInventive |
http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06T3-40
http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06T17-00
http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06T9-00
http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06N3-08
http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06T5-50 |
| filingDate |
2021-01-12-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
| inventor |
http://rdf.ncbi.nlm.nih.gov/pubchem/patentinventor/MD5_752d322ae6e3cafe743b8edd44339ec3
http://rdf.ncbi.nlm.nih.gov/pubchem/patentinventor/MD5_9f58ec6a8d5ef96e94bb7786d33aef8e |
| publicationDate |
2021-05-04-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
| publicationNumber |
CN-112750198-A |
| titleOfInvention |
A Dense Correspondence Prediction Method Based on Nonrigid Point Clouds |
| abstract |
The invention discloses a dense correspondence prediction method based on non-rigid point cloud, comprising: extracting geometric features of three-dimensional template and point cloud by using graph convolutional neural network and multiple set abstraction layers; The associated global features of the cloud infer the global displacement; use the local feature embedding technology and introduce an attention mechanism to fuse the local depth features of the point cloud with the geometric features of the graph; use the local regression network to predict the displacement increment; use a weakly supervised fine-tuning method , the real point cloud is processed and unified with the two-stage regression network in a complete framework. The invention not only makes full use of the local geometric features of the point cloud, adopts the attention strategy to improve the corresponding accuracy, but also adopts the weak supervision fine-tuning method to process the real point cloud robustly, which effectively improves the unreasonable distortion of the prediction model caused by the lack of training data, and the input Conspicuous inconsistencies in shape. |
| isCitedBy |
http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-114091628-B
http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-114091628-A
http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-113487728-A
http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-114913552-A |
| priorityDate |
2021-01-12-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
| type |
http://data.epo.org/linked-data/def/patent/Publication |