EventGPT: Event Stream Understanding with Multimodal Large Language Models

Event cameras record visual information as asynchronous pixel change streams, excelling at scene perception under unsatisfactory lighting or high-dynamic conditions. Existing multimodal large language models (MLLMs) concentrate on natural RGB images, failing in scenarios where event data fits better. In this paper, we introduce EventGPT, the first MLLM for event stream understanding, to the best of our knowledge, marking a pioneering attempt to integrate large language models (LLMs) with event stream comprehension. To mitigate the huge domain gaps, we develop a three-stage optimization paradigm to gradually equip a pre-trained LLM with the capability of understanding event-based scenes. Our EventGPT comprises an event encoder, followed by a spatio-temporal aggregator, a linear projector, an event-language adapter, and an LLM. Firstly, RGB image-text pairs generated by GPT are leveraged to warm up the linear projector, referring to LLaVA, as the gap between natural image and language modalities is relatively smaller. Secondly, we construct a synthetic yet large dataset, N-ImageNet-Chat, consisting of event frames and corresponding texts to enable the use of the spatio-temporal aggregator and to train the event-language adapter, thereby aligning event features more closely with the language space. Finally, we gather an instruction dataset, Event-Chat, which contains extensive real-world data to fine-tune the entire model, further enhancing its generalization ability. We construct a comprehensive benchmark, and experiments show that EventGPT surpasses previous state-of-the-art MLLMs in generation quality, descriptive accuracy, and reasoning capability.

RAIL: Robot Affordance Imagination with Large Language Models

This paper introduces an automatic affordance reasoning paradigm tailored to minimal semantic inputs, addressing the critical challenges of classifying and manipulating unseen classes of objects in household settings. Inspired by human cognitive processes, our method integrates generative language models and physics-based simulators to foster analytical thinking and creative imagination of novel affordances. Structured with a tripartite framework consisting of analysis, imagination, and evaluation, our system "analyzes" the requested affordance names into interaction-based definitions, "imagines" the virtual scenarios, and "evaluates" the object affordance. If an object is recognized as possessing the requested affordance, our method also predicts the optimal pose for such functionality, and how a potential user can interact with it. Tuned on only a few synthetic examples across 3 affordance classes, our pipeline achieves a very high success rate on affordance classification and functional pose prediction of 8 classes of novel objects, outperforming learning-based baselines. Validation through real robot manipulating experiments demonstrates the practical applicability of the imagined user interaction, showcasing the system's ability to independently conceptualize unseen affordances and interact with new objects and scenarios in everyday settings.

High-speed Low-consumption sEMG-based Transient-state micro-Gesture Recognition

Gesture recognition on wearable devices is extensively applied in human-computer interaction. Electromyography (EMG) has been used in many gesture recognition systems for its rapid perception of muscle signals. However, analyzing EMG signals on devices, like smart wristbands, usually needs inference models to have high performances, such as low inference latency, low power consumption, and low memory occupation. Therefore, this paper proposes an improved spiking neural network (SNN) to achieve these goals. We propose an adaptive multi-delta coding as a spiking coding method to improve recognition accuracy. We propose two additive solvers for SNN, which can reduce inference energy consumption and amount of parameters significantly, and improve the robustness of temporal differences. In addition, we propose a linear action detection method TAD-LIF, which is suitable for SNNs. TAD-LIF is an improved LIF neuron that can detect transient-state gestures quickly and accurately. We collected two datasets from 20 subjects including 6 micro gestures. The collection devices are two designed lightweight consumer-level sEMG wristbands (3 and 8 electrode channels respectively). Compared to CNN, FCN, and normal SNN-based methods, the proposed SNN has higher recognition accuracy. The accuracy of the proposed SNN is 83.85% and 93.52% on the two datasets respectively. In addition, the inference latency of the proposed SNN is about 1% of CNN, the power consumption is about 0.1% of CNN, and the memory occupation is about 20% of CNN. The proposed methods can be used for precise, high-speed, and low-power micro-gesture recognition tasks, and are suitable for consumer-level intelligent wearable devices, which is a general way to achieve ubiquitous computing.

Prepare the Chair for the Bear! Robot Imagination of Sitting Affordance to Reorient Previously Unseen Chairs

In this letter, a paradigm for the classification and manipulation of previously unseen objects is established and demonstrated through a real example of chairs. We present a novel robot manipulation method, guided by the understanding of object stability, perceptibility, and affordance, which allows the robot to prepare previously unseen and randomly oriented chairs for a teddy bear to sit on. Specifically, the robot encounters an unknown object and first reconstructs a complete 3D model from perceptual data via active and autonomous manipulation. By inserting this model into a physical simulator (i.e., the robot's "imagination"), the robot assesses whether the object is a chair and determines how to reorient it properly to be used, i.e., how to reorient it to an upright and accessible pose. If the object is classified as a chair, the robot reorients the object to this pose and seats the teddy bear onto the chair. The teddy bear is a proxy for an elderly person, hospital patient, or child. Experiment results show that our method achieves a high success rate on the real robot task of chair preparation. Also, it outperforms several baseline methods on the task of upright pose prediction for chairs.

PRIMP: PRobabilistically-Informed Motion Primitives for Efficient Affordance Learning from Demonstration

This paper proposes a learning-from-demonstration method using probability densities on the workspaces of robot manipulators. The method, named "PRobabilistically-Informed Motion Primitives (PRIMP)", learns the probability distribution of the end effector trajectories in the 6D workspace that includes both positions and orientations. It is able to adapt to new situations such as novel via poses with uncertainty and a change of viewing frame. The method itself is robot-agnostic, in which the learned distribution can be transferred to another robot with the adaptation to its workspace density. The learned trajectory distribution is then used to guide an optimization-based motion planning algorithm to further help the robot avoid novel obstacles that are unseen during the demonstration process. The proposed methods are evaluated by several sets of benchmark experiments. PRIMP runs more than 5 times faster while generalizing trajectories more than twice as close to both the demonstrations and novel desired poses. It is then combined with our robot imagination method that learns object affordances, illustrating the applicability of PRIMP to learn tool use through physical experiments.

Pink-Eggs Dataset V1: A Step Toward Invasive Species Management Using Deep Learning Embedded Solutions

We introduce a novel dataset consisting of images depicting pink eggs that have been identified as Pomacea canaliculata eggs, accompanied by corresponding bounding box annotations. The purpose of this dataset is to aid researchers in the analysis of the spread of Pomacea canaliculata species by utilizing deep learning techniques, as well as supporting other investigative pursuits that require visual data pertaining to the eggs of Pomacea canaliculata. It is worth noting, however, that the identity of the eggs in question is not definitively established, as other species within the same taxonomic family have been observed to lay similar-looking eggs in regions of the Americas. Therefore, a crucial prerequisite to any decision regarding the elimination of these eggs would be to establish with certainty whether they are exclusively attributable to invasive Pomacea canaliculata or if other species are also involved. The dataset is available at https://www.kaggle.com/datasets/deeshenzhen/pinkeggs

Peeling the Onion: Hierarchical Reduction of Data Redundancy for Efficient Vision Transformer Training

Vision transformers (ViTs) have recently obtained success in many applications, but their intensive computation and heavy memory usage at both training and inference time limit their generalization. Previous compression algorithms usually start from the pre-trained dense models and only focus on efficient inference, while time-consuming training is still unavoidable. In contrast, this paper points out that the million-scale training data is redundant, which is the fundamental reason for the tedious training. To address the issue, this paper aims to introduce sparsity into data and proposes an end-to-end efficient training framework from three sparse perspectives, dubbed Tri-Level E-ViT. Specifically, we leverage a hierarchical data redundancy reduction scheme, by exploring the sparsity under three levels: number of training examples in the dataset, number of patches (tokens) in each example, and number of connections between tokens that lie in attention weights. With extensive experiments, we demonstrate that our proposed technique can noticeably accelerate training for various ViT architectures while maintaining accuracy. Remarkably, under certain ratios, we are able to improve the ViT accuracy rather than compromising it. For example, we can achieve 15.2% speedup with 72.6% (+0.4) Top-1 accuracy on Deit-T, and 15.7% speedup with 79.9% (+0.1) Top-1 accuracy on Deit-S. This proves the existence of data redundancy in ViT.

HeatViT: Hardware-Efficient Adaptive Token Pruning for Vision Transformers

While vision transformers (ViTs) have continuously achieved new milestones in the field of computer vision, their sophisticated network architectures with high computation and memory costs have impeded their deployment on resource-limited edge devices. In this paper, we propose a hardware-efficient image-adaptive token pruning framework called HeatViT for efficient yet accurate ViT acceleration on embedded FPGAs. By analyzing the inherent computational patterns in ViTs, we first design an effective attention-based multi-head token selector, which can be progressively inserted before transformer blocks to dynamically identify and consolidate the non-informative tokens from input images. Moreover, we implement the token selector on hardware by adding miniature control logic to heavily reuse existing hardware components built for the backbone ViT. To improve the hardware efficiency, we further employ 8-bit fixed-point quantization, and propose polynomial approximations with regularization effect on quantization error for the frequently used nonlinear functions in ViTs. Finally, we propose a latency-aware multi-stage training strategy to determine the transformer blocks for inserting token selectors and optimize the desired (average) pruning rates for inserted token selectors, in order to improve both the model accuracy and inference latency on hardware. Compared to existing ViT pruning studies, under the similar computation cost, HeatViT can achieve 0.7%$\sim$8.9% higher accuracy; while under the similar model accuracy, HeatViT can achieve more than 28.4%$\sim$65.3% computation reduction, for various widely used ViTs, including DeiT-T, DeiT-S, DeiT-B, LV-ViT-S, and LV-ViT-M, on the ImageNet dataset. Compared to the baseline hardware accelerator, our implementations of HeatViT on the Xilinx ZCU102 FPGA achieve 3.46$\times$$\sim$4.89$\times$ speedup.

The Lottery Ticket Hypothesis for Vision Transformers

The conventional lottery ticket hypothesis (LTH) claims that there exists a sparse subnetwork within a dense neural network and a proper random initialization method, called the winning ticket, such that it can be trained from scratch to almost as good as the dense counterpart. Meanwhile, the research of LTH in vision transformers (ViTs) is scarcely evaluated. In this paper, we first show that the conventional winning ticket is hard to find at weight level of ViTs by existing methods. Then, we generalize the LTH for ViTs to input images consisting of image patches inspired by the input dependence of ViTs. That is, there exists a subset of input image patches such that a ViT can be trained from scratch by using only this subset of patches and achieve similar accuracy to the ViTs trained by using all image patches. We call this subset of input patches the winning tickets, which represent a significant amount of information in the input. Furthermore, we present a simple yet effective method to find the winning tickets in input patches for various types of ViT, including DeiT, LV-ViT, and Swin Transformers. More specifically, we use a ticket selector to generate the winning tickets based on the informativeness of patches. Meanwhile, we build another randomly selected subset of patches for comparison, and the experiments show that there is clear difference between the performance of models trained with winning tickets and randomly selected subsets.

Transporters with Visual Foresight for Solving Unseen Rearrangement Tasks

Rearrangement tasks have been identified as a crucial challenge for intelligent robotic manipulation, but few methods allow for precise construction of unseen structures. We propose a visual foresight model for pick-and-place manipulation which is able to learn efficiently. In addition, we develop a multi-modal action proposal module which builds on Goal-Conditioned Transporter Networks, a state-of-the-art imitation learning method. Our method, Transporters with Visual Foresight (TVF), enables task planning from image data and is able to achieve multi-task learning and zero-shot generalization to unseen tasks with only a handful of expert demonstrations. TVF is able to improve the performance of a state-of-the-art imitation learning method on both training and unseen tasks in simulation and real robot experiments. In particular, the average success rate on unseen tasks improves from 55.0% to 77.9% in simulation experiments and from 30% to 63.3% in real robot experiments when given only tens of expert demonstrations. More details can be found on our project website: https://chirikjianlab.github.io/tvf/