Data Fusion for Cross-Domain Real-Time Object Detection on the Edge
We investigate an edge-computing scenario for robot control, where two similar neural networks are running on one computational node. We test the feasibility of using a single object-detection model (YOLOv5) with the benefit of reduced computational resources against the potentially more accurate in...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-07-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/23/13/6138 |
| _version_ | 1797590776697323520 |
|---|---|
| author | Mykyta Kovalenko David Przewozny Peter Eisert Sebastian Bosse Paul Chojecki |
| author_facet | Mykyta Kovalenko David Przewozny Peter Eisert Sebastian Bosse Paul Chojecki |
| author_sort | Mykyta Kovalenko |
| collection | DOAJ |
| description | We investigate an edge-computing scenario for robot control, where two similar neural networks are running on one computational node. We test the feasibility of using a single object-detection model (YOLOv5) with the benefit of reduced computational resources against the potentially more accurate independent and specialized models. Our results show that using one single convolutional neural network (for object detection and hand-gesture classification) instead of two separate ones can reduce resource usage by almost <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>50</mn><mo>%</mo></mrow></semantics></math></inline-formula>. For many classes, we observed an increase in accuracy when using the model trained with more labels. For small datasets (a few hundred instances per label), we found that it is advisable to add labels with many instances from another dataset to increase detection accuracy. |
| first_indexed | 2024-03-11T01:28:13Z |
| format | Article |
| id | doaj.art-471be74226a2486f84faf8b61c356d76 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-11T01:28:13Z |
| publishDate | 2023-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-471be74226a2486f84faf8b61c356d762023-11-18T17:31:05ZengMDPI AGSensors1424-82202023-07-012313613810.3390/s23136138Data Fusion for Cross-Domain Real-Time Object Detection on the EdgeMykyta Kovalenko0David Przewozny1Peter Eisert2Sebastian Bosse3Paul Chojecki4Fraunhofer Heinrich Hertz Institute, 10587 Berlin, GermanyFraunhofer Heinrich Hertz Institute, 10587 Berlin, GermanyFraunhofer Heinrich Hertz Institute, 10587 Berlin, GermanyFraunhofer Heinrich Hertz Institute, 10587 Berlin, GermanyFraunhofer Heinrich Hertz Institute, 10587 Berlin, GermanyWe investigate an edge-computing scenario for robot control, where two similar neural networks are running on one computational node. We test the feasibility of using a single object-detection model (YOLOv5) with the benefit of reduced computational resources against the potentially more accurate independent and specialized models. Our results show that using one single convolutional neural network (for object detection and hand-gesture classification) instead of two separate ones can reduce resource usage by almost <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>50</mn><mo>%</mo></mrow></semantics></math></inline-formula>. For many classes, we observed an increase in accuracy when using the model trained with more labels. For small datasets (a few hundred instances per label), we found that it is advisable to add labels with many instances from another dataset to increase detection accuracy.https://www.mdpi.com/1424-8220/23/13/6138object detectionedge computinghuman-computer interactionvisual analysisoptimization |
| spellingShingle | Mykyta Kovalenko David Przewozny Peter Eisert Sebastian Bosse Paul Chojecki Data Fusion for Cross-Domain Real-Time Object Detection on the Edge Sensors object detection edge computing human-computer interaction visual analysis optimization |
| title | Data Fusion for Cross-Domain Real-Time Object Detection on the Edge |
| title_full | Data Fusion for Cross-Domain Real-Time Object Detection on the Edge |
| title_fullStr | Data Fusion for Cross-Domain Real-Time Object Detection on the Edge |
| title_full_unstemmed | Data Fusion for Cross-Domain Real-Time Object Detection on the Edge |
| title_short | Data Fusion for Cross-Domain Real-Time Object Detection on the Edge |
| title_sort | data fusion for cross domain real time object detection on the edge |
| topic | object detection edge computing human-computer interaction visual analysis optimization |
| url | https://www.mdpi.com/1424-8220/23/13/6138 |
| work_keys_str_mv | AT mykytakovalenko datafusionforcrossdomainrealtimeobjectdetectionontheedge AT davidprzewozny datafusionforcrossdomainrealtimeobjectdetectionontheedge AT petereisert datafusionforcrossdomainrealtimeobjectdetectionontheedge AT sebastianbosse datafusionforcrossdomainrealtimeobjectdetectionontheedge AT paulchojecki datafusionforcrossdomainrealtimeobjectdetectionontheedge |