Masked Image Modeling (MIM) has emerged as a powerful self-supervised learning paradigm for visual representation learning, enabling models to acquire rich visual representations by predicting masked portions of images from their visible regions. While this approach has shown promising results, we hypothesize that its effectiveness may be limited by optimization challenges during early training stages, where models are expected to learn complex image distributions from partial observations before developing basic visual processing capabilities. To address this limitation, we propose a prototype-driven curriculum learning framework that structures the learning process to progress from prototypical examples to more complex variations in the dataset. Our approach introduces a temperature-based annealing scheme that gradually expands the training distribution, enabling more stable and efficient learning trajectories. Through extensive experiments on ImageNet-1K, we demonstrate that our curriculum learning strategy significantly improves both training efficiency and representation quality while requiring substantially fewer training epochs compared to standard Masked Auto-Encoding. Our findings suggest that carefully controlling the order of training examples plays a crucial role in self-supervised visual learning, providing a practical solution to the early-stage optimization challenges in MIM.

Contrastive image-text pre-trained models such as CLIP have shown remarkable adaptability to downstream tasks. However, they face challenges due to the high computational requirements of the Vision Transformer (ViT) backbone. Current strategies to boost ViT efficiency focus on pruning patch tokens but fall short in addressing the multimodal nature of CLIP and identifying the optimal subset of tokens for maximum performance. To address this, we propose greedy search methods to establish a “Golden Ranking” and introduce a lightweight predictor specifically trained to approximate this Ranking. To compensate for any performance degradation resulting from token pruning, we incorporate learnable visual tokens that aid in restoring and potentially enhancing the model's performance. Our work presents a comprehensive and systematic investigation of pruning tokens within the ViT backbone of CLIP models. Through our framework, we successfully reduced 40% of patch tokens in CLIP's ViT while only suffering a minimal average accuracy loss of 0.3% across seven datasets. Our study lays the groundwork for building more computationally efficient multimodal models without sacrificing their performance, addressing a key challenge in the application of advanced vision-language models.11Project Page: https://github.com/CEWu/PatchRanking

Our work tackles the computational challenges of contrastive learning methods, particularly for the pretraining of Vision Transformers (ViTs). Despite the effectiveness of contrastive learning, the substantial computational resources required for training often hinder their practical application. To mitigate this issue, we propose an acceleration framework, leveraging ViT's unique ability to generalize across inputs of varying sequence lengths. Our method employs a mix of sequence compression strategies, including randomized token dropout and flexible patch scaling, to reduce the cost of gradient estimation and accelerate convergence. We further provide an in-depth analysis of the gradient estimation error of various acceleration strategies as well as their impact on downstream tasks, offering valuable insights into the trade-offs between acceleration and performance. We also propose a novel procedure to identify an optimal acceleration schedule to adjust the sequence compression ratios to the training progress, ensuring efficient training without sacrificing downstream performance. Our approach significantly reduces computational overhead across various self-supervised learning algorithms on large-scale datasets. In ImageNet, our method achieves speedups of 4$\times$ in MoCo, 3.3$\times$ in SimCLR, and 2.5$\times$ in DINO, demonstrating substantial efficiency gains.

Vision-language models such as CLIP [27] learn a generic text-image embedding from large-scale training data. A vision-language model can be adapted to a new classification task through few-shot prompt tuning. We find that such a prompt tuning process is highly robust to label noises. This intrigues us to study the key reasons contributing to the robustness of the prompt tuning paradigm. We conducted extensive experiments to explore this property and find the key factors are: 1) the fixed classname tokens provide a strong regularization to the optimization of the model, reducing gradients induced by the noisy samples; 2) the powerful pre-trained image-text embedding that is learned from diverse and generic web data provides strong prior knowledge for image classification. Further, we demonstrate that noisy zero-shot predictions from CLIP can be used to tune its own prompt, significantly enhancing prediction accuracy in the unsupervised setting. The code is available at https://github.com/CEWu/PTNL.

Automated Guided Vehicles (AGVs) can be widely used in many applications. In a dynamic environment where humans and machines coexist, it is expected to encounter traffic jams, queuing, overtaking, comity, and many other diverse, dynamic social interaction. However, capturing the characteristics of the social interactions among passages for path planning is a complex problem. It is crucial to realize the application of AGVs. Dealing with such a complex dynamic environment thus becomes a primary subject of navigation. However, in facing stop-and-go crowds in shopping malls or traffic jams, the past navigation methods cannot effectively cope with the navigation task. Therefore, we propose a hierarchical reinforcement learning framework named MTON, which collects historic environmental statistics to analyze social interactions and periodically selects a promising navigation subgoal to complete the navigation. Our model introduces the idea of interchanging between waiting and moving. Thus it can cope with stop-and-go dynamic obstacles and find a way to break through the barriers to the final goal.

Counting the number of people is a common and basic computing operation in many applications. Most of the people counting techniques need a sensing device like camera and apply image processing methods to track pedestrians. However, counting people with cameras in private places raises a lot of security and privacy issues. The passive infra-red sensor (PIR) can detect the body temperature of the infrared and thus provides another promising solution. Although a single PIR can easily identify the passing situations (i.e., in or out) of a single person, the signals of a single PIR is not sufficient to identify the complex situations of multiple people. In the paper, we design a people counting device with a PIR array to detect the passing situations and generate data records with higher discriminability. In addition, we apply the machine learning classification methods including the CNN, the RBM+LR, Decision Tree, and NaiveBayes on the collected data records to identify the passing situations. To validate our design, we conduct experiments to study the feasibility and classification performance and explore the impact factors. The experimental results show that the CNN outperforms the other and achieves the best accuracy, i.e., about 92%. Also, the results show that the captured data records of the PIR array contain sufficient characteristics for identifying complex passing situations and the configuration of the PIR array including the sensor direction and the field of view (FOV) of a PIR modified by the metal tape can significantly impact the discriminability of the collected data.

It is hard to detect on-road objects under various lighting conditions. To improve the quality of the classifier, three techniques are used. We define subclasses to separate daytime and nighttime samples. Then we skip similar samples in the training set to prevent overfitting. With the help of the outside training samples, the detection accuracy is also improved. To detect objects in an edge device, Nvidia Jetson TX2 platform, we exert the lightweight model ResNet-18 FPN as the backbone feature extractor. The FPN (Feature Pyramid Network) generates good features for detecting objects over various scales. With Cascade R-CNN technique, the bounding boxes are iteratively refined for better results.

Continual lifelong learning is essential to many applications. In this paper, we propose a simple but effective approach to continual deep learning. Our approach leverages the principles of deep model compression, critical weights selection, and progressive networks expansion. By enforcing their integration in an iterative manner, we introduce an incremental learning method that is scalable to the number of sequential tasks in a continual learning process. Our approach is easy to implement and owns several favorable characteristics. First, it can avoid forgetting (i.e., learn new tasks while remembering all previous tasks). Second, it allows model expansion but can maintain the model compactness when handling sequential tasks. Besides, through our compaction and selection/expansion mechanism, we show that the knowledge accumulated through learning previous tasks is helpful to build a better model for the new tasks compared to training the models independently with tasks. Experimental results show that our approach can incrementally learn a deep model tackling multiple tasks without forgetting, while the model compactness is maintained with the performance more satisfiable than individual task training.