QPILOTS: Efficient Test-Time Q-Steering for Flow Policies

Flow-matching and diffusion policies are expressive action generators, but optimizing them with temporal-difference reinforcement learning (RL) remains difficult. Effective policy extraction requires exploiting the critic's action gradient, yet directly backpropagating this signal through a multi-step denoising process can be numerically unstable. Existing methods work around this either by discarding gradient information, distilling the policy into a simpler one-step actor, or repeatedly fine-tuning the denoising policy as the critic improves. We propose QPILOTS, a method that leaves the original policy unmodified and steers the denoising process at inference time. At each denoising step, instead of evaluating the critic on the noisy intermediate action where critic predictions are unreliable, we first project that intermediate state to an estimate of the final clean action and compute the critic gradient there. We introduce two variants: QPILOTS-U uses a fast single-point approximation, while QPILOTS-M draws differentiable posterior samples via a learned auxiliary network. On a standard offline-to-online RL benchmark, QPILOTS achieves the best aggregate performance, reaching an average success rate of 90% across 50 tasks. We also apply QPILOTS to steer a large, frozen, pretrained Vision-Language Action (VLA) foundation model, outperforming or matching prior inference-time approaches across six manipulation tasks in simulation.

Dual-Triplet Metric Learning for Unsupervised Domain Adaptation in Video-Based Face Recognition

The scalability and complexity of deep learning models remains a key issue in many of visual recognition applications like, e.g., video surveillance, where fine tuning with labeled image data from each new camera is required to reduce the domain shift between videos captured from the source domain, e.g., a laboratory setting, and the target domain, i.e, an operational environment. In many video surveillance applications, like face recognition (FR) and person re-identification, a pair-wise matcher is used to assign a query image captured using a video camera to the corresponding reference images in a gallery. The different configurations and operational conditions of video cameras can introduce significant shifts in the pair-wise distance distributions, resulting in degraded recognition performance for new cameras. In this paper, a new deep domain adaptation (DA) method is proposed to adapt the CNN embedding of a Siamese network using unlabeled tracklets captured with a new video cameras. To this end, a dual-triplet loss is introduced for metric learning, where two triplets are constructed using video data from a source camera, and a new target camera. In order to constitute the dual triplets, a mutual-supervised learning approach is introduced where the source camera acts as a teacher, providing the target camera with an initial embedding. Then, the student relies on the teacher to iteratively label the positive and negative pairs collected during, e.g., initial camera calibration. Both source and target embeddings continue to simultaneously learn such that their pair-wise distance distributions become aligned. For validation, the proposed metric learning technique is used to train deep Siamese networks under different training scenarios, and is compared to state-of-the-art techniques for still-to-video FR on the COX-S2V and a private video-based FR dataset.

Cross-Domain Face Synthesis using a Controllable GAN

The performance of face recognition (FR) systems applied in video surveillance has been shown to improve when the design data is augmented through synthetic face generation. This is true, for instance, with pair-wise matchers (e.g., deep Siamese networks) that typically rely on a reference gallery with one still image per individual. However, generating synthetic images in the source domain may not improve the performance during operations due to the domain shift w.r.t. the target domain. Moreover, despite the emergence of Generative Adversarial Networks (GANs) for realistic synthetic generation, it is often difficult to control the conditions under which synthetic faces are generated. In this paper, a cross-domain face synthesis approach is proposed that integrates a new Controllable GAN (C-GAN). It employs an off-the-shelf 3D face model as a simulator to generate face images under various poses. The simulated images and noise are input to the C-GAN for realism refinement which employs an additional adversarial game as a third player to preserve the identity and specific facial attributes of the refined images. This allows generating realistic synthetic face images that reflects capture conditions in the target domain while controlling the GAN output to generate faces under desired pose conditions. Experiments were performed using videos from the Chokepoint and COX-S2V datasets, and a deep Siamese network for FR with a single reference still per person. Results indicate that the proposed approach can provide a higher level of accuracy compared to the current state-of-the-art approaches for synthetic data augmentation.