MiniNet: a concise CNN for image forgery detection

Yadav, Divakar

doi:10.1007/s12530-022-09446-0

Cited by 8 publications

(6 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…noise inconsistencies [6,8,15], double JPEG compression [16,17], etc. These approaches are useful for tampering, for example, copymove, splicing, etc.…”

Section: Rel Ated Workmentioning

confidence: 99%

“…This progress has encouraged researchers to develop novel deep neural network models for image forensics. Recent works based on CNNs [17][18][19][20], recurrent neural networks [4,21], and transformers [22][23][24] perform much better for Forgery detection than traditional methods.…”

Section: Rel Ated Workmentioning

confidence: 99%

“…It also includes tasks such as the detection and localization of tampered regions. Traditional Image forensics methods use hand‐crafted features such as CFA artifacts [7], noise inconsistencies [6,8,15], double JPEG compression [16,17], etc. These approaches are useful for tampering, for example, copy‐move, splicing, etc.…”

Section: Related Workmentioning

confidence: 99%

See 2 more Smart Citations

ForensicNet: Modern convolutional neural network‐based image forgery detection network

Yadav

2023

Journal of Forensic Sciences

Self Cite

View full text Add to dashboard Cite

The advancements in Image editing techniques produce realistic‐looking artificial images with ease. These images can easily circumvent the forensic systems making the authentication process more tedious and difficult. To overcome this problem, we introduce a modern convolutional neural network (CNN) named ForensicNet, inspired by the recent developments in computer vision. The three major contributions of our CNNs are inverted bottleneck, separate downsampling layers, and using depth‐wise convolutions for mixing information in the spatial dimension. The inverted bottlenecks help improve accuracy and reduce network parameters/FLOPs. The separate downsampling layers help converge the network. The normalization layers also help stabilize training whenever the spatial resolution is changed. The depth‐wise convolution is a grouped convolution where the number of groups and channels are the same. The experiments show that ForensicNet outperforms the state‐of‐the‐art methods by a large margin.

show abstract

“…noise inconsistencies [6,8,15], double JPEG compression [16,17], etc. These approaches are useful for tampering, for example, copymove, splicing, etc.…”

Section: Rel Ated Workmentioning

confidence: 99%

Section: Rel Ated Workmentioning

confidence: 99%

See 1 more Smart Citation

ForensicNet: Modern convolutional neural network‐based image forgery detection network

Yadav

2023

Journal of Forensic Sciences

Self Cite

View full text Add to dashboard Cite

show abstract

“…At the forefront are deep learning models, renowned for their proficiency in pattern recognition, which have been fine-tuned to identify the subtlest of anomalies characteristic of fabricated images. These sophisticated models scrutinize for peculiarities in elements such as texture, lighting, and edge delineation-attributes that typically betray an image's synthetic origins and may elude the human eye's detection (Tahaoglu et al, 2022;Tanaka et al, 2021;Tyagi & Yadav, 2022). In addition to visual analysis, some strategies delve into the digital DNA of images by examining their metadata or tracing the origins of their distribution (Wu et al, 2022).…”

Section: Introductionmentioning

confidence: 99%

“…This ongoing contest not only fuels technological advancement but also raises the stakes in ensuring the integrity of shared media content. The relentless advancement in AI technology propels the task of preserving content authenticity into an ever-evolving challenge (Tyagi & Yadav, 2022). Social media platforms are finding themselves at the crux of this issue, tasked with the responsibility of safeguarding user trust and the integrity of the content shared within their domains.…”

Section: Introductionmentioning

confidence: 99%

Untitled

2024

GMJ

View full text Add to dashboard Cite

The proliferation of images generated by artificial intelligence (AI) has significantly impacted the digital landscape, especially on social media platforms where the distinction between natural and synthetic content is increasingly blurred. This study embarks on a comparative review of the strategies used by major social media platforms-Facebook/Instagram, Twitter, TikTok, and YouTube-to detect AI-generated images. Employing a comprehensive methodology that includes a systematic review of academic literature, analysis of platform policies, and expert interviews, this research assesses the effectiveness of various detection methods, ranging from sophisticated AI tools to user reporting mechanisms. The findings reveal diverse approaches: Facebook and Instagram utilise a blend of AI detection and human moderation; Twitter integrates machine learning algorithms with user reports; TikTok emphasises AI tools within moderation workflows and educational initiatives; and YouTube relies on its Content ID system alongside AI analysis. The study highlights the critical role of effective detection systems in maintaining content authenticity and user trust, underscoring the importance of balancing automated detection with human oversight. The ongoing development and refinement of these technologies, alongside collaborative efforts and evolving regulatory frameworks, are identified as essential for ensuring a trustworthy digital environment. This research contributes to the discourse on digital integrity, offering insights into the complexities of safeguarding social media ecosystems against the challenges posed by AI-generated content.

show abstract