Er mwyn gwireddu deallusrwydd artiffisial, mae'n rhaid i ni integreiddio deallusrwydd peirianyddol a deallusrwydd dynol. Mae’r broses o ddelweddu yn defnyddio cynrychioliadau gweledol rhyngweithiol o ddata i wella gwybyddiaeth trwy ddefnyddio galluoedd canfyddiadol y llygad dynol. Yn oes data mawr, mae'n chwarae rhan hanfodol gan ei fod yn ein helpu i wneud synnwyr o lawer iawn o wybodaeth gymhleth. Mae dadansoddeg weledol yn defnyddio technegau a rhyngwynebau delweddu i greu amgylchedd dadansoddol sy’n ymdrochol. Mae hyn yn ein galluogi i ddefnyddio dealltwriaeth weledol a dealltwriaeth beiriannol fel dull cyflenwol er mwyn cyfuno deallusrwydd dynol a pheiriannol. Mewn senarios dadansoddi a gwneud penderfyniadau sy'n canolbwyntio ar bobl, megis diogelwch, milwrol, ac atal a lliniaru trychinebau, bydd dadansoddi gweledol rhyngweithiol a deallus yn dod yn ddull dadansoddol craidd.
Nodau
- Gwneud gwaith ymchwil dwys ym maes technegau delweddu a dadansoddeg weledol
- Dod i hyd i gymwysiadau rhyngddisgyblaethol ar gyfer delweddu a dadansoddeg weledol
Ymchwil
Technegau delweddu cyffredinol
- Graffeg Gyfrifiadurol
- Animeiddio a rendro
- Prosesu delweddau a signalau
- Dylunio, modelu a phrosesu geometrig
- Realiti Estynedig (AR)/Realiti Rhithwir (VR) / arddangosiadau ymdrochol
Sylfeini mathemategol ac algorithmau dysgu peirianyddol
- Clystyru a chyfuno data
- Lleihau dimensiynoldeb
- Darganfod, echdynnu, olrhain a thrawsnewid nodweddion.
- Technegau dysgu peirianyddol
- Dulliau rhifiadol
Meysydd lle gallwch chi ddefnyddio delweddu a dadansoddeg gweledol
- Dysgu Peirianyddol
- Iechyd a Chymdeithas
- Peirianneg a Roboteg
- Y Gwyddorau Cymdeithasol
Prosiectau cyfredol
- Datgelu "greddfau" modelau cynhyrchiol dwfn at ddibenion creu cynnwys gweledol sy’n deg a diduedd. Cyllid o tua £72,922. Partneriaethau Hyfforddiant Doethurol EPSRC
Cwrdd â’r tîm
Digwyddiadau
Cyflwynir seminarau gan aelodau ac ymwelwyr yn y gyfres seminarau ymchwil cyfrifiadura gweledol.
Cyhoeddiadau
- Yang, W. et al., 2024. Interactive reweighting for mitigating label quality issues. IEEE Transactions on Visualization and Computer Graphics 30 (3), pp.1837-1852. (10.1109/TVCG.2023.3345340)
- Huang, Z. et al., 2024. SpeechMirror: A multimodal visual analytics system for personalized reflection of online public speaking effectiveness. IEEE Transactions on Visualization and Computer Graphics 30 (1), pp.606-616. (10.1109/TVCG.2023.3326932)
- Chen, C. et al., 2023. A unified interactive model evaluation for classification, object detection, and instance segmentation in computer vision. Presented at: IEEE VIS 2023 Melbourne, Australia 22-27 October 2023. Institute of Electrical and Electronics Engineers(10.1109/TVCG.2023.3326588)
- Song, S. et al. 2023. Feature proliferation — the "cancer" in StyleGAN and its treatments. Presented at: International Conference on Computer Vision (ICCV) 2023 Paris, France October 1 - 6, 2023. Proceedings of IEEE/CVF International Conference on Computer Vision. IEEE. , pp.2360-2370. (10.1109/ICCV51070.2023.00224)
- Deng, Z. et al. 2023. Sketch2PQ: freeform planar quadrilateral mesh design via a single sketch. IEEE Transactions on Visualization and Computer Graphics 29 (9), pp.3826-3839. (10.1109/TVCG.2022.3170853)
- Xia, J. et al., 2023. Interactive visual cluster analysis by contrastive dimensionality reduction. IEEE Transactions on Visualization and Computer Graphics 29 (1), pp.734-744. (10.1109/TVCG.2022.3209423)
- Li, Z. et al., 2022. A unified understanding of deep NLP models for text classification. IEEE Transactions on Visualization and Computer Graphics 28 (12), pp.4980-4994. (10.1109/TVCG.2022.3184186)
- Grange, J. A. et al. 2022. XAI & I: Self-explanatory AI facilitating mutual understanding between AI and human experts. Presented at: 26th International Conference on Knowledge-Based and Intelligent Information & Engineering Systems (KES 2022) 7-9 September 2022. Elsevier(10.1016/j.procs.2022.09.419)
- Chen, C. et al., 2022. Towards better caption supervision for object detection. IEEE Transactions on Visualization and Computer Graphics 28 (4), pp.1941-1954. (10.1109/TVCG.2021.3138933)
- Maher, K. et al., 2022. E-ffective: a visual analytic system for exploring the emotion andeffectiveness of inspirational speeches. IEEE Transactions on Visualization and Computer Graphics 28 (1), pp.508-517. (10.1109/TVCG.2021.3114789)
- Liang, Y. et al. 2022. Exploring and exploiting hubness priors for high-quality GAN latent sampling. Presented at: The 39th International Conference on Machine Learning (ICML 2022) Baltimore, Maryland USA 17-23 July 2022. Vol. 162.
- Chen, C. et al., 2021. Interactive graph construction for graph-based semi-supervised learning. IEEE Transactions on Visualization and Computer Graphics 27 (9), pp.3701-3716. (10.1109/TVCG.2021.3084694)
- Cao, K. et al., 2021. Analyzing the noise robustness of deep neural networks. IEEE Transactions on Visualization and Computer Graphics 27 (7), pp.3289-3304. (10.1109/TVCG.2020.2969185)
- Pan, Y. et al., 2019. InSocialNet: Interactive visual analytics for role-event videos. Computational Visual Media 5 (4), pp.375-390. (10.1007/s41095-019-0157-9)
- Zhang, J. et al., 2019. Accelerating ADMM for efficient simulation and optimization. ACM Transactions on Graphics 38 (6) 163. (10.1145/3355089.3356491)
- Xiang, S. et al., 2019. Interactive correction of mislabeled training data. Presented at: IEEE VIS 2019 Vancouver, BC, Canada 20-25 October 2019. 2019 IEEE Conference on Visual Analytics Science and Technology (VAST). IEEE(10.1109/VAST47406.2019.8986943)
- Zhang, J. et al., 2019. Static/dynamic filtering for mesh geometry. IEEE Transactions on Visualization and Computer Graphics 25 (4), pp.1774-1787. (10.1109/TVCG.2018.2816926)
- Liu, S. et al., 2018. Visual diagnosis of tree boosting methods. IEEE Transactions on Visualization and Computer Graphics 24 (1), pp.163-173. (10.1109/TVCG.2017.2744378)
Y camau nesaf
Ymchwil sy’n gwneud gwahaniaeth
Mae ein hymchwil yn gwneud gwahaniaeth i fywydau pobl wrth i ni gweithio ar draws disgyblaethau er mwyn ymgodymu â phrif heriau sy’n wynebu’r gymdeithas, yr economi ac ein hamgylchedd.
Ymchwil ôl-raddedig
Mae ein graddau ymchwil yn rhoi'r rhyddid i chi i archwilio pwnc arbennig mewn dyfnder ymhlith ymchwilwyr blaenllaw.
Ein heffaith ymchwil
Mae'r astudiaethau achos hyn yn rhoi sylw i rai o'r meysydd lle rydym yn cael effaith ymchwil gadarnhaol.