Suspect Face Generation

Jalan, Harsh Jaykumar; Maurya, Gautam; Corda, Canute; Dsouza, Sunny; Panchal, Dakshata

doi:10.1109/cscita47329.2020.9137812

Cited by 4 publications

(1 citation statement)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It would potentially improve information retrieval to the extent of making it easier for more accurate facial images to be obtained when searching the web, and more importantly, it would make software and web browsing a dramatically better experience for users with visual impairment (Makav and Kılıc ¸, 2019). It is also helpful in forensic analysis (Jalan et al, 2020), bridging the gap between face descriptions and what those faces actually look like. This also affects the work being done in the inverse task of generating facial images from descriptions, which would lend the power of artificial intelligence to the work currently being done by (computer-aided) sketch artists.…”

Section: Introductionmentioning

confidence: 99%

Face2Text revisited: Improved data set and baseline results

Tanti¹,

Abdilla²,

Muscat³

et al. 2022

Preprint

View full text Add to dashboard Cite

Current image description generation models do not transfer well to the task of describing human faces. To encourage the development of more human-focused descriptions, we developed a new data set of facial descriptions based on the CelebA image data set. We describe the properties of this data set, and present results from a face description generator trained on it, which explores the feasibility of using transfer learning from VGGFace/ResNet CNNs. Comparisons are drawn through both automated metrics and human evaluation by 76 English-speaking participants. The descriptions generated by the VGGFace-LSTM + Attention model are closest to the ground truth according to human evaluation whilst the ResNet-LSTM + Attention model obtained the highest CIDEr and CIDEr-D results (1.252 and 0.686 respectively). Together, the new data set and these experimental results provide data and baselines for future work in this area.

show abstract

Section: Introductionmentioning

confidence: 99%