Browser fingerprinting is used for bot detection. In response, bots have started altering their fingerprints to evade detection. We conduct the first large-scale evaluation to study whether and how altering fingerprints helps bots evade detection. To systematically investigate such evasive bots, we deploy a honey site that includes two anti-bot services (DataDome and BotD) and solicit bot traffic from 20 different bot services that purport to sell ''realistic and undetectable traffic.'' Across half a million requests recorded on our honey site, we find an average evasion rate of 52.93% against DataDome and 44.56% evasion rate against BotD. Our analysis of fingerprint attributes of evasive bots shows that they indeed alter their fingerprints. Moreover, we find that the attributes of these altered fingerprints are often inconsistent with each other. We propose FP-Inconsistent, a data-driven approach to detect such inconsistencies across space (two attributes in a given browser fingerprint) and time (a single attribute at two different points in time). Our evaluation shows that our approach can reduce the evasion rate of evasive bots by 44.95%-48.11% while maintaining a true negative rate of 96.84% on traffic from real users.

In this paper, we present a method for robust head pose estimation via carefully designed loss functions. We propose that exploiting the relationship between the predicted yaw, pitch, roll values and the features of a head pose estimation network is crucial to make robust predictions. With the aim of achieving the above criteria, we formulate novel loss functions that assure robustness and generalization of the network predictions. We report results on public datasets namely, AFLW2000-3D and BIWI demonstrating that the proposed method outperforms the state-of-the-art 2-d head pose estimation algorithms by a margin of up to 10%. We will release the source code at https://github.com/soni-H/lercpose/ upon acceptance of the paper.

Generalized zero-shot action recognition is a challenging problem, where the task is to recognize new action categories that are unavailable during the training stage, in addition to the seen action categories. Existing approaches suffer from the inherent bias of the learned classifier towards the seen action categories. As a consequence, unseen category samples are incorrectly classified as belonging to one of the seen action categories. In this paper, we set out to tackle this issue by arguing for a separate treatment of seen and unseen action categories in generalized zero-shot action recognition. We introduce an out-of-distribution detector that determines whether the video features belong to a seen or unseen action category. To train our out-of-distribution detector, video features for unseen action categories are synthesized using generative adversarial networks trained on seen action category features. To the best of our knowledge, we are the first to propose an out-of-distribution detector based GZSL framework for action recognition in videos. Experiments are performed on three action recognition datasets: Olympic Sports, HMDB51 and UCF101. For generalized zero-shot action recognition, our proposed approach outperforms the baseline with absolute gains (in classification accuracy) of 7.0%, 3.4%, and 4.9%, respectively, on these datasets.