Provable Ordering and Continuity in Vision-Language Pretraining for Generalizable Embodied Agents

Pre-training vision-language representations on human action videos has emerged as a promising approach to reduce reliance on large-scale expert demonstrations for training embodied agents. However, prior methods often employ time contrastive learning based on goal-reaching heuristics, progressively aligning language instructions from the initial to the final frame. This overemphasis on future frames can result in erroneous vision-language associations, as actions may terminate early or include irrelevant moments in the end. To address this issue, we propose Action Temporal Coherence Learning (AcTOL) to learn ordered and continuous vision-language representations without rigid goal-based constraint. AcTOL treats a video as a continuous trajectory where it (1) contrasts semantic differences between frames to reflect their natural ordering, and (2) imposes a local Brownian bridge constraint to ensure smooth transitions across intermediate frames. Extensive imitation learning experiments across varying numbers of demonstrations show that the pretrained features significantly enhance downstream manipulation tasks by up to 49% with high robustness to different linguistic styles of instructions, offering a viable pathway toward generalized embodied agents. The source code is included in the supplementary material for reference.

PolaFormer: Polarity-aware Linear Attention for Vision Transformers

Linear attention has emerged as a promising alternative to softmax-based attention, leveraging kernelized feature maps to reduce complexity from quadratic to linear in sequence length. However, the non-negative constraint on feature maps and the relaxed exponential function used in approximation lead to significant information loss compared to the original query-key dot products, resulting in less discriminative attention maps with higher entropy. To address the missing interactions driven by negative values in query-key pairs, we propose a polarity-aware linear attention mechanism that explicitly models both same-signed and opposite-signed query-key interactions, ensuring comprehensive coverage of relational information. Furthermore, to restore the spiky properties of attention maps, we provide a theoretical analysis proving the existence of a class of element-wise functions (with positive first and second derivatives) that can reduce entropy in the attention distribution. For simplicity, and recognizing the distinct contributions of each dimension, we employ a learnable power function for rescaling, allowing strong and weak attention signals to be effectively separated. Extensive experiments demonstrate that the proposed PolaFormer improves performance on various vision tasks, enhancing both expressiveness and efficiency by up to 4.6%.

Color-Oriented Redundancy Reduction in Dataset Distillation

Dataset Distillation (DD) is designed to generate condensed representations of extensive image datasets, enhancing training efficiency. Despite recent advances, there remains considerable potential for improvement, particularly in addressing the notable redundancy within the color space of distilled images. In this paper, we propose AutoPalette, a framework that minimizes color redundancy at the individual image and overall dataset levels, respectively. At the image level, we employ a palette network, a specialized neural network, to dynamically allocate colors from a reduced color space to each pixel. The palette network identifies essential areas in synthetic images for model training and consequently assigns more unique colors to them. At the dataset level, we develop a color-guided initialization strategy to minimize redundancy among images. Representative images with the least replicated color patterns are selected based on the information gain. A comprehensive performance study involving various datasets and evaluation scenarios is conducted, demonstrating the superior performance of our proposed color-aware DD compared to existing DD methods. The code is available at \url{https://github.com/KeViNYuAn0314/AutoPalette}.

Track Any Peppers: Weakly Supervised Sweet Pepper Tracking Using VLMs

In the Detection and Multi-Object Tracking of Sweet Peppers Challenge, we present Track Any Peppers (TAP) - a weakly supervised ensemble technique for sweet peppers tracking. TAP leverages the zero-shot detection capabilities of vision-language foundation models like Grounding DINO to automatically generate pseudo-labels for sweet peppers in video sequences with minimal human intervention. These pseudo-labels, refined when necessary, are used to train a YOLOv8 segmentation network. To enhance detection accuracy under challenging conditions, we incorporate pre-processing techniques such as relighting adjustments and apply depth-based filtering during post-inference. For object tracking, we integrate the Matching by Segment Anything (MASA) adapter with the BoT-SORT algorithm. Our approach achieves a HOTA score of 80.4%, MOTA of 66.1%, Recall of 74.0%, and Precision of 90.7%, demonstrating effective tracking of sweet peppers without extensive manual effort. This work highlights the potential of foundation models for efficient and accurate object detection and tracking in agricultural settings.

Tokens on Demand: Token Condensation as Training-free Test-time Adaptation

In this work, we introduce Token Condensation as Adaptation (TCA), a training-free approach designed to mitigate distribution shifts encountered by vision-language models (VLMs) during test-time inference. TCA bridges distribution gaps at the patch level by condensing image tokens that exhibit low attentiveness to the <cls> token. Recognizing the <cls> token may correspond to universal concepts, TCA identifies and tracks the most reliable <cls> tokens that align specifically with target classes from historical data streams. To achieve this, we propose a context token reservoir (CTR), which retains tokens with the lowest uncertainty as ``anchors" to guide the preservation of class-relevant tokens during inference. These anchors, in turn, act as token-level classifiers to correct VLM predictions and improve visual-text alignment. Utilizing anchors sampled from CTR, TCA condenses tokens through two operations: (1) pruning class-irrelevant tokens that consistently rank low across all attention heads to reach cross-head consensus on their irrelevance, and (2) merging the remaining class-ambiguous tokens into representative centers using coreset selection, maintaining linear computational complexity. As the first method to explore token efficiency in test-time adaptation, TCA consistently demonstrates superior performance across cross-dataset and out-of-distribution adaptation tasks, reducing GFLOPs by 12.2% to 48.9% while achieving accuracy improvements up to 21.4% against the strongest baseline without introducing additional parameters.

CF-PRNet: Coarse-to-Fine Prototype Refining Network for Point Cloud Completion and Reconstruction

In modern agriculture, precise monitoring of plants and fruits is crucial for tasks such as high-throughput phenotyping and automated harvesting. This paper addresses the challenge of reconstructing accurate 3D shapes of fruits from partial views, which is common in agricultural settings. We introduce CF-PRNet, a coarse-to-fine prototype refining network, leverages high-resolution 3D data during the training phase but requires only a single RGB-D image for real-time inference. Our approach begins by extracting the incomplete point cloud data that constructed from a partial view of a fruit with a series of convolutional blocks. The extracted features inform the generation of scaling vectors that refine two sequentially constructed 3D mesh prototypes - one coarse and one fine-grained. This progressive refinement facilitates the detailed completion of the final point clouds, achieving detailed and accurate reconstructions. CF-PRNet demonstrates excellent performance metrics with a Chamfer Distance of 3.78, an F1 Score of 66.76%, a Precision of 56.56%, and a Recall of 85.31%, and win the first place in the Shape Completion and Reconstruction of Sweet Peppers Challenge.

OC3D: Weakly Supervised Outdoor 3D Object Detection with Only Coarse Click Annotation

LiDAR-based outdoor 3D object detection has received widespread attention. However, training 3D detectors from the LiDAR point cloud typically relies on expensive bounding box annotations. This paper presents OC3D, an innovative weakly supervised method requiring only coarse clicks on the bird' s eye view of the 3D point cloud. A key challenge here is the absence of complete geometric descriptions of the target objects from such simple click annotations. To address this problem, our proposed OC3D adopts a two-stage strategy. In the first stage, we initially design a novel dynamic and static classification strategy and then propose the Click2Box and Click2Mask modules to generate box-level and mask-level pseudo-labels for static and dynamic instances, respectively. In the second stage, we design a Mask2Box module, leveraging the learning capabilities of neural networks to update mask-level pseudo-labels, which contain less information, to box level pseudo-labels. Experimental results on the widely used KITTI and nuScenes datasets demonstrate that our OC3D with only coarse clicks achieves state-of-the-art performance compared to weakly-supervised 3D detection methods. Combining OC3D with a missing click mining strategy, we propose a OC3D++ pipeline, which requires only 0.2% annotation cost in the KITTI dataset to achieve performance comparable to fully supervised methods.

FastEdit: Fast Text-Guided Single-Image Editing via Semantic-Aware Diffusion Fine-Tuning

Conventional Text-guided single-image editing approaches require a two-step process, including fine-tuning the target text embedding for over 1K iterations and the generative model for another 1.5K iterations. Although it ensures that the resulting image closely aligns with both the input image and the target text, this process often requires 7 minutes per image, posing a challenge for practical application due to its time-intensive nature. To address this bottleneck, we introduce FastEdit, a fast text-guided single-image editing method with semantic-aware diffusion fine-tuning, dramatically accelerating the editing process to only 17 seconds. FastEdit streamlines the generative model's fine-tuning phase, reducing it from 1.5K to a mere 50 iterations. For diffusion fine-tuning, we adopt certain time step values based on the semantic discrepancy between the input image and target text. Furthermore, FastEdit circumvents the initial fine-tuning step by utilizing an image-to-image model that conditions on the feature space, rather than the text embedding space. It can effectively align the target text prompt and input image within the same feature space and save substantial processing time. Additionally, we apply the parameter-efficient fine-tuning technique LoRA to U-net. With LoRA, FastEdit minimizes the model's trainable parameters to only 0.37\% of the original size. At the same time, we can achieve comparable editing outcomes with significantly reduced computational overhead. We conduct extensive experiments to validate the editing performance of our approach and show promising editing capabilities, including content addition, style transfer, background replacement, and posture manipulation, etc.

MOS: Model Synergy for Test-Time Adaptation on LiDAR-Based 3D Object Detection

LiDAR-based 3D object detection is pivotal across many applications, yet the performance of such detection systems often degrades after deployment, especially when faced with unseen test point clouds originating from diverse locations or subjected to corruption. In this work, we introduce a new online adaptation framework for detectors named Model Synergy (MOS). Specifically, MOS dynamically assembles best-fit supermodels for each test batch from a bank of historical checkpoints, leveraging long-term knowledge to guide model updates without forgetting. The model assembly is directed by the proposed synergy weights (SW), employed for weighted averaging of the selected checkpoints to minimize redundancy in the composite supermodel. These weights are calculated by evaluating the similarity of predicted bounding boxes on test data and the feature independence among model pairs in the bank. To maintain an informative yet compact model bank, we pop out checkpoints with the lowest average SW scores and insert newly updated model weights. Our method was rigorously tested against prior test-time domain adaptation strategies on three datasets and under eight types of corruptions, demonstrating its superior adaptability to changing scenes and conditions. Remarkably, our approach achieved a 67.3% increase in performance in a complex "cross-corruption" scenario, which involves cross-dataset inconsistencies and real-world scene corruptions, providing a more realistic testbed of adaptation capabilities. The code is available at https://github.com/zhuoxiao-chen/MOS.

DiPEx: Dispersing Prompt Expansion for Class-Agnostic Object Detection

Class-agnostic object detection (OD) can be a cornerstone or a bottleneck for many downstream vision tasks. Despite considerable advancements in bottom-up and multi-object discovery methods that leverage basic visual cues to identify salient objects, consistently achieving a high recall rate remains difficult due to the diversity of object types and their contextual complexity. In this work, we investigate using vision-language models (VLMs) to enhance object detection via a self-supervised prompt learning strategy. Our initial findings indicate that manually crafted text queries often result in undetected objects, primarily because detection confidence diminishes when the query words exhibit semantic overlap. To address this, we propose a Dispersing Prompt Expansion (DiPEx) approach. DiPEx progressively learns to expand a set of distinct, non-overlapping hyperspherical prompts to enhance recall rates, thereby improving performance in downstream tasks such as out-of-distribution OD. Specifically, DiPEx initiates the process by self-training generic parent prompts and selecting the one with the highest semantic uncertainty for further expansion. The resulting child prompts are expected to inherit semantics from their parent prompts while capturing more fine-grained semantics. We apply dispersion losses to ensure high inter-class discrepancy among child prompts while preserving semantic consistency between parent-child prompt pairs. To prevent excessive growth of the prompt sets, we utilize the maximum angular coverage (MAC) of the semantic space as a criterion for early termination. We demonstrate the effectiveness of DiPEx through extensive class-agnostic OD and OOD-OD experiments on MS-COCO and LVIS, surpassing other prompting methods by up to 20.1% in AR and achieving a 21.3% AP improvement over SAM. The code is available at https://github.com/jason-lim26/DiPEx.