The precise age estimation of child sexual abuse and exploitation (CSAE) victims is one of the most significant digital forensic challenges. Investigators often need to determine the age of victims by looking at images and interpreting the sexual development stages and other human characteristics. The main priority - safeguarding children -- is often negatively impacted by a huge forensic backlog, cognitive bias and the immense psychological stress that this work can entail. This paper evaluates existing facial image datasets and proposes a new dataset tailored to the needs of similar digital forensic research contributions. This small, diverse dataset of 0 to 20-year-old individuals contains 245 images and is merged with 82 unique images from the FG-NET dataset, thus achieving a total of 327 images with high image diversity and low age range density. The new dataset is tested on the Deep EXpectation (DEX) algorithm pre-trained on the IMDB-WIKI dataset. The overall results for young adolescents aged 10 to 15 and older adolescents/adults aged 16 to 20 are very encouraging -- achieving MAEs as low as 1.79, but also suggest that the accuracy for children aged 0 to 10 needs further work. In order to determine the efficacy of the prototype, valuable input of four digital forensic experts, including two forensic investigators, has been taken into account to improve age estimation results. Further research is required to extend datasets both concerning image density and the equal distribution of factors such as gender and racial diversity.
Gray-box hyperparameter optimization techniques have recently emerged as a promising direction for tuning Deep Learning methods. In this work, we introduce DyHPO, a method that learns to dynamically decide which configuration to try next, and for what budget. Our technique is a modification to the classical Bayesian optimization for a gray-box setup. Concretely, we propose a new surrogate for Gaussian Processes that embeds the learning curve dynamics and a new acquisition function that incorporates multi-budget information. We demonstrate the significant superiority of DyHPO against state-of-the-art hyperparameter optimization baselines through large-scale experiments comprising 50 datasets (Tabular, Image, NLP) and diverse neural networks (MLP, CNN/NAS, RNN).
The ground truth used for training image, video, or speech quality prediction models is based on the Mean Opinion Scores (MOS) obtained from subjective experiments. Usually, it is necessary to conduct multiple experiments, mostly with different test participants, to obtain enough data to train quality models based on machine learning. Each of these experiments is subject to an experiment-specific bias, where the rating of the same file may be substantially different in two experiments (e.g. depending on the overall quality distribution). These different ratings for the same distortion levels confuse neural networks during training and lead to lower performance. To overcome this problem, we propose a bias-aware loss function that estimates each dataset's biases during training with a linear function and considers it while optimising the network weights. We prove the efficiency of the proposed method by training and validating quality prediction models on synthetic and subjective image and speech quality datasets.
Nuclei segmentation is a fundamental task in digital pathology analysis and can be automated by deep learning-based methods. However, the development of such an automated method requires a large amount of data with precisely annotated masks which is hard to obtain. Training with weakly labeled data is a popular solution for reducing the workload of annotation. In this paper, we propose a novel meta-learning-based nuclei segmentation method which follows the label correction paradigm to leverage data with noisy masks. Specifically, we design a fully conventional meta-model that can correct noisy masks using a small amount of clean meta-data. Then the corrected masks can be used to supervise the training of the segmentation model. Meanwhile, a bi-level optimization method is adopted to alternately update the parameters of the main segmentation model and the meta-model in an end-to-end way. Extensive experimental results on two nuclear segmentation datasets show that our method achieves the state-of-the-art result. It even achieves comparable performance with the model training on supervised data in some noisy settings.
Machines that can represent and describe environmental soundscapes have practical potential, e.g., for audio tagging and captioning systems. Prevailing learning paradigms have been relying on parallel audio-text data, which is, however, scarcely available on the web. We propose VIP-ANT that induces \textbf{A}udio-\textbf{T}ext alignment without using any parallel audio-text data. Our key idea is to share the image modality between bi-modal image-text representations and bi-modal image-audio representations; the image modality functions as a pivot and connects audio and text in a tri-modal embedding space implicitly. In a difficult zero-shot setting with no paired audio-text data, our model demonstrates state-of-the-art zero-shot performance on the ESC50 and US8K audio classification tasks, and even surpasses the supervised state of the art for Clotho caption retrieval (with audio queries) by 2.2\% R@1. We further investigate cases of minimal audio-text supervision, finding that, e.g., just a few hundred supervised audio-text pairs increase the zero-shot audio classification accuracy by 8\% on US8K. However, to match human parity on some zero-shot tasks, our empirical scaling experiments suggest that we would need about $2^{21} \approx 2M$ supervised audio-caption pairs. Our work opens up new avenues for learning audio-text connections with little to no parallel audio-text data.
Remote sensing change detection, identifying changes between scenes of the same location, is an active area of research with a broad range of applications. Recent advances in multimodal self-supervised pretraining have resulted in state-of-the-art methods which surpass vision models trained solely on optical imagery. In the remote sensing field, there is a wealth of overlapping 2D and 3D modalities which can be exploited to supervise representation learning in vision models. In this paper we propose Contrastive Surface-Image Pretraining (CSIP) for joint learning using optical RGB and above ground level (AGL) map pairs. We then evaluate these pretrained models on several building segmentation and change detection datasets to show that our method does, in fact, extract features relevant to downstream applications where natural and artificial surface information is relevant.
Video classification systems are vulnerable to adversarial attacks, which can create severe security problems in video verification. Current black-box attacks need a large number of queries to succeed, resulting in high computational overhead in the process of attack. On the other hand, attacks with restricted perturbations are ineffective against defenses such as denoising or adversarial training. In this paper, we focus on unrestricted perturbations and propose StyleFool, a black-box video adversarial attack via style transfer to fool the video classification system. StyleFool first utilizes color theme proximity to select the best style image, which helps avoid unnatural details in the stylized videos. Meanwhile, the target class confidence is additionally considered in targeted attack to influence the output distribution of the classifier by moving the stylized video closer to or even across the decision boundary. A gradient-free method is then employed to further optimize the adversarial perturbation. We carry out extensive experiments to evaluate StyleFool on two standard datasets, UCF-101 and HMDB-51. The experimental results suggest that StyleFool outperforms the state-of-the-art adversarial attacks in terms of both number of queries and robustness against existing defenses. We identify that 50% of the stylized videos in untargeted attack do not need any query since they can already fool the video classification model. Furthermore, we evaluate the indistinguishability through a user study to show that the adversarial samples of StyleFool look imperceptible to human eyes, despite unrestricted perturbations.
Despite data augmentation being a de facto technique for boosting the performance of deep neural networks, little attention has been paid to developing augmentation strategies for generative adversarial networks (GANs). To this end, we introduce a novel augmentation scheme designed specifically for GAN-based semantic image synthesis models. We propose to randomly warp object shapes in the semantic label maps used as an input to the generator. The local shape discrepancies between the warped and non-warped label maps and images enable the GAN to learn better the structural and geometric details of the scene and thus to improve the quality of generated images. While benchmarking the augmented GAN models against their vanilla counterparts, we discover that the quantification metrics reported in the previous semantic image synthesis studies are strongly biased towards specific semantic classes as they are derived via an external pre-trained segmentation network. We therefore propose to improve the established semantic image synthesis evaluation scheme by analyzing separately the performance of generated images on the biased and unbiased classes for the given segmentation network. Finally, we show strong quantitative and qualitative improvements obtained with our augmentation scheme, on both class splits, using state-of-the-art semantic image synthesis models across three different datasets. On average across COCO-Stuff, ADE20K and Cityscapes datasets, the augmented models outperform their vanilla counterparts by ~3 mIoU and ~10 FID points.
Recent advances in unsupervised representation learning have experienced remarkable progress, especially with the achievements of contrastive learning, which regards each image as well its augmentations as a separate class, while does not consider the semantic similarity among images. This paper proposes a new kind of data augmentation, named Center-wise Local Image Mixture, to expand the neighborhood space of an image. CLIM encourages both local similarity and global aggregation while pulling similar images. This is achieved by searching local similar samples of an image, and only selecting images that are closer to the corresponding cluster center, which we denote as center-wise local selection. As a result, similar representations are progressively approaching the clusters, while do not break the local similarity. Furthermore, image mixture is used as a smoothing regularization to avoid overconfidence on the selected samples. Besides, we introduce multi-resolution augmentation, which enables the representation to be scale invariant. Integrating the two augmentations produces better feature representation on several unsupervised benchmarks. Notably, we reach 75.5% top-1 accuracy with linear evaluation over ResNet-50, and 59.3% top-1 accuracy when fine-tuned with only 1% labels, as well as consistently outperforming supervised pretraining on several downstream transfer tasks.
Adversarial patches are optimized contiguous pixel blocks in an input image that cause a machine-learning model to misclassify it. However, their optimization is computationally demanding, and requires careful hyperparameter tuning, potentially leading to suboptimal robustness evaluations. To overcome these issues, we propose ImageNet-Patch, a dataset to benchmark machine-learning models against adversarial patches. It consists of a set of patches, optimized to generalize across different models, and readily applicable to ImageNet data after preprocessing them with affine transformations. This process enables an approximate yet faster robustness evaluation, leveraging the transferability of adversarial perturbations. We showcase the usefulness of this dataset by testing the effectiveness of the computed patches against 127 models. We conclude by discussing how our dataset could be used as a benchmark for robustness, and how our methodology can be generalized to other domains. We open source our dataset and evaluation code at https://github.com/pralab/ImageNet-Patch.