Instruction tuning large language model (LLM) on image-text pairs has achieved unprecedented vision-language multimodal abilities. However, their vision-language alignments are only built on image-level, the lack of region-level alignment limits their advancements to fine-grained multimodal understanding. In this paper, we propose instruction tuning on region-of-interest. The key design is to reformulate the bounding box as the format of spatial instruction. The interleaved sequences of visual features extracted by the spatial instruction and the language embedding are input to LLM, and trained on the transformed region-text data in instruction tuning format. Our region-level vision-language model, termed as GPT4RoI, brings brand new conversational and interactive experience beyond image-level understanding. (1) Controllability: Users can interact with our model by both language and spatial instructions to flexibly adjust the detail level of the question. (2) Capacities: Our model supports not only single-region spatial instruction but also multi-region. This unlocks more region-level multimodal capacities such as detailed region caption and complex region reasoning. (3) Composition: Any off-the-shelf object detector can be a spatial instruction provider so as to mine informative object attributes from our model, like color, shape, material, action, relation to other objects, etc. The code, data, and demo can be found at https://github.com/jshilong/GPT4RoI.
The missing signal caused by the objects being occluded or an unstable sensor is a common challenge during data collection. Such missing signals will adversely affect the results obtained from the data, and this issue is observed more frequently in robotic tactile perception. In tactile perception, due to the limited working space and the dynamic environment, the contact between the tactile sensor and the object is frequently insufficient and unstable, which causes the partial loss of signals, thus leading to incomplete tactile data. The tactile data will therefore contain fewer tactile cues with low information density. In this paper, we propose a tactile representation learning method, named TacMAE, based on Masked Autoencoder to address the problem of incomplete tactile data in tactile perception. In our framework, a portion of the tactile image is masked out to simulate the missing contact region. By reconstructing the missing signals in the tactile image, the trained model can achieve a high-level understanding of surface geometry and tactile properties from limited tactile cues. The experimental results of tactile texture recognition show that our proposed TacMAE can achieve a high recognition accuracy of 71.4% in the zero-shot transfer and 85.8% after fine-tuning, which are 15.2% and 8.2% higher than the results without using masked modeling. The extensive experiments on YCB objects demonstrate the knowledge transferability of our proposed method and the potential to improve efficiency in tactile exploration.
Indonesia, as a maritime country, has a significant portion of its territory covered by water. Ineffective waste management has resulted in a considerable amount of trash in Indonesian waters, leading to various issues. The development of an automated trash-collecting robot can be a solution to address this problem. The robot requires a system capable of detecting objects in motion, such as in videos. However, using naive object detection methods in videos has limitations, particularly when image focus is reduced and the target object is obstructed by other objects. This paper's contribution provides an explanation of the methods that can be applied to perform video object detection in an automated trash-collecting robot. The study utilizes the YOLOv5 model and the Robust & Efficient Post Processing (REPP) method, along with tubelet-level bounding box linking on the FloW and Roboflow datasets. The combination of these methods enhances the performance of naive object detection from YOLOv5 by considering the detection results in adjacent frames. The results show that the post-processing stage and tubelet-level bounding box linking can improve the quality of detection, achieving approximately 3% better performance compared to YOLOv5 alone. The use of these methods has the potential to detect surface and underwater trash and can be applied to a real-time image-based trash-collecting robot. Implementing this system is expected to mitigate the damage caused by trash in the past and improve Indonesia's waste management system in the future.
Dual energy cargo inspection systems are sensitive to both the area density and the atomic number of an imaged container due to the Z dependence of photon attenuation. The ability to identify cargo contents by their atomic number enables improved detection capabilities of illicit materials. Existing methods typically classify materials into a few material classes using an empirical calibration step. However, such a coarse label discretization limits atomic number selectivity and can yield inaccurate results if a material is near the midpoint of two bins. This work introduces a high resolution atomic number prediction method by minimizing the chi-squared error between measured transparency values and a semiempirical transparency model. Our previous work showed that by incorporating calibration step, the semiempirical transparency model can capture second order effects such as scattering. This method is benchmarked using two simulated radiographic phantoms, demonstrating the ability to obtain accurate material predictions on noisy input images by incorporating an image segmentation step. Furthermore, we show that this approach can be adapted to identify shielded objects after first determining the properties of the shielding, taking advantage of the closed-form nature of the transparency model.
We propose an image restoration algorithm that can control the perceptual quality and/or the mean square error (MSE) of any pre-trained model, trading one over the other at test time. Our algorithm is few-shot: Given about a dozen images restored by the model, it can significantly improve the perceptual quality and/or the MSE of the model for newly restored images without further training. Our approach is motivated by a recent theoretical result that links between the minimum MSE (MMSE) predictor and the predictor that minimizes the MSE under a perfect perceptual quality constraint. Specifically, it has been shown that the latter can be obtained by optimally transporting the output of the former, such that its distribution matches the source data. Thus, to improve the perceptual quality of a predictor that was originally trained to minimize MSE, we approximate the optimal transport by a linear transformation in the latent space of a variational auto-encoder, which we compute in closed-form using empirical means and covariances. Going beyond the theory, we find that applying the same procedure on models that were initially trained to achieve high perceptual quality, typically improves their perceptual quality even further. And by interpolating the results with the original output of the model, we can improve their MSE on the expense of perceptual quality. We illustrate our method on a variety of degradations applied to general content images of arbitrary dimensions.
This work addresses the challenging domain adaptation setting in which knowledge from the labelled source domain dataset is available only from the pretrained black-box segmentation model. The pretrained model's predictions for the target domain images are noisy because of the distributional differences between the source domain data and the target domain data. Since the model's predictions serve as pseudo labels during self-training, the noise in the predictions impose an upper bound on model performance. Therefore, we propose a simple yet novel image translation workflow, ReGEN, to address this problem. ReGEN comprises an image-to-image translation network and a segmentation network. Our workflow generates target-like images using the noisy predictions from the original target domain images. These target-like images are semantically consistent with the noisy model predictions and therefore can be used to train the segmentation network. In addition to being semantically consistent with the predictions from the original target domain images, the generated target-like images are also stylistically similar to the target domain images. This allows us to leverage the stylistic differences between the target-like images and the target domain image as an additional source of supervision while training the segmentation model. We evaluate our model with two benchmark domain adaptation settings and demonstrate that our approach performs favourably relative to recent state-of-the-art work. The source code will be made available.
Deep learning has become a valuable tool for the automation of certain medical image segmentation tasks, significantly relieving the workload of medical specialists. Some of these tasks require segmentation to be performed on a subset of the input dimensions, the most common case being 3D-to-2D. However, the performance of existing methods is strongly conditioned by the amount of labeled data available, as there is currently no data efficient method, e.g. transfer learning, that has been validated on these tasks. In this work, we propose a novel convolutional neural network (CNN) and self-supervised learning (SSL) method for label-efficient 3D-to-2D segmentation. The CNN is composed of a 3D encoder and a 2D decoder connected by novel 3D-to-2D blocks. The SSL method consists of reconstructing image pairs of modalities with different dimensionality. The approach has been validated in two tasks with clinical relevance: the en-face segmentation of geographic atrophy and reticular pseudodrusen in optical coherence tomography. Results on different datasets demonstrate that the proposed CNN significantly improves the state of the art in scenarios with limited labeled data by up to 8% in Dice score. Moreover, the proposed SSL method allows further improvement of this performance by up to 23%, and we show that the SSL is beneficial regardless of the network architecture.
Smartphones equipped with a multi-camera system comprising multiple cameras with different field-of-view (FoVs) are becoming more prevalent. These camera configurations are compatible with reference-based SR and video SR, which can be executed simultaneously while recording video on the device. Thus, combining these two SR methods can improve image quality. Recently, Lee et al. have presented such a method, RefVSR. In this paper, we consider how to optimally utilize the observations obtained, including input low-resolution (LR) video and reference (Ref) video. RefVSR extends conventional video SR quite simply, aggregating the LR and Ref inputs over time in a single bidirectional stream. However, considering the content difference between LR and Ref images due to their FoVs, we can derive the maximum information from the two image sequences by aggregating them independently in the temporal direction. Then, we propose an improved method, RefVSR++, which can aggregate two features in parallel in the temporal direction, one for aggregating the fused LR and Ref inputs and the other for Ref inputs over time. Furthermore, we equip RefVSR++ with enhanced mechanisms to align image features over time, which is the key to the success of video SR. We experimentally show that RefVSR++ outperforms RefVSR by over 1dB in PSNR, achieving the new state-of-the-art.
We evaluate existing foundation models video understanding capabilities using a carefully designed experiment protocol consisting of three hallmark tasks (action recognition, temporal localization, and spatiotemporal localization), eight datasets well received by the community, and four adaptation methods tailoring a foundation model (FM) for a downstream task. Moreover, we propose a scalar VideoGLUE score (VGS) to measure an FMs efficacy and efficiency when adapting to general video understanding tasks. Our main findings are as follows. First, task-specialized models significantly outperform the six FMs studied in this work, in sharp contrast to what FMs have achieved in natural language and image understanding. Second,video-native FMs, whose pretraining data contains the video modality, are generally better than image-native FMs in classifying motion-rich videos, localizing actions in time, and understanding a video of more than one action. Third, the video-native FMs can perform well on video tasks under light adaptations to downstream tasks(e.g., freezing the FM backbones), while image-native FMs win in full end-to-end finetuning. The first two observations reveal the need and tremendous opportunities to conduct research on video-focused FMs, and the last confirms that both tasks and adaptation methods matter when it comes to the evaluation of FMs.
Recently, U-shaped networks have dominated the field of medical image segmentation due to their simple and easily tuned structure. However, existing U-shaped segmentation networks: 1) mostly focus on designing complex self-attention modules to compensate for the lack of long-term dependence based on convolution operation, which increases the overall number of parameters and computational complexity of the network; 2) simply fuse the features of encoder and decoder, ignoring the connection between their spatial locations. In this paper, we rethink the above problem and build a lightweight medical image segmentation network, called SegNetr. Specifically, we introduce a novel SegNetr block that can perform local-global interactions dynamically at any stage and with only linear complexity. At the same time, we design a general information retention skip connection (IRSC) to preserve the spatial location information of encoder features and achieve accurate fusion with the decoder features. We validate the effectiveness of SegNetr on four mainstream medical image segmentation datasets, with 59\% and 76\% fewer parameters and GFLOPs than vanilla U-Net, while achieving segmentation performance comparable to state-of-the-art methods. Notably, the components proposed in this paper can be applied to other U-shaped networks to improve their segmentation performance.