In the recent decade, high-throughput plant phenotyping techniques, which combine non-invasive image analysis and machine learning, have been successfully applied to identify and quantify plant health and diseases. However, these techniques usually do not consider the progressive nature of plant stress and often require images showing severe signs of stress to ensure high confidence detection, thereby reducing the feasibility for early detection and recovery of plants under stress. To overcome the problem mentioned above, we propose a deep learning pipeline for the temporal analysis of the visual changes induced in the plant due to stress and apply it to the specific water stress identification case in Chickpea plant shoot images. For this, we have considered an image dataset of two chickpea varieties JG-62 and Pusa-372, under three water stress conditions; control, young seedling, and before flowering, captured over five months. We have employed a variant of Convolutional Neural Network - Long Short Term Memory (CNN-LSTM) network to learn spatio-temporal patterns from the chickpea plant dataset and use them for water stress classification. Our model has achieved ceiling level classification performance of 98.52% on JG-62 and 97.78% on Pusa-372 chickpea plant data and has outperformed the best reported time-invariant technique by at least 14% for both JG-62 and Pusa-372 species, to the best of our knowledge. Furthermore, our CNN-LSTM model has demonstrated robustness to noisy input, with a less than 2.5% dip in average model accuracy and a small standard deviation about the mean for both species. Lastly, we have performed an ablation study to analyze the performance of the CNN-LSTM model by decreasing the number of temporal session data used for training.
카메라와 비전 센서는 제조 환경에서 제품 및 공정 검사에 널리 사용된다. 전통적으로 이러한 모델을 학습시키려면 정상 제품과 불량 제품 모두에 대해 수천 장의 이미지가 필요했다. 최근 연구에 따르면 CLIP과 같은 비전 트랜스포머(ViT) 모델은 이 요구량을 정상 및 불량 클래스당 50~100장으로 줄일 수 있다. 그러나 수율이 높거나 생산량이 적은 제조 공정에서는 불량 이미지 50장을 수집하는 것조차 어려울 수 있다. 본 논문에서는 최근 제안된 두 구조화 의사결정 모델(Jev와 Clef-Flash)의 유용성을, 추론 시 클래스당 참조 이미지를 k=2장까지 적게 사용하는 응용에서 살펴본다. 이 과정에서 우리는 세 가지 주요 기여를 제시한다. (a) 이미지 분류/검사 응용을 위한 이러한 모델의 k-샷 학습 프레임워크, (b) 이러한 의사결정 모델에서 이미지 분류를 안내하는 텍스트 설명을 자동으로 개선하는 에이전트적 접근법, (c) 검사 의사결정의 신뢰성을 높이기 위해 결과로 얻은 결함 확률을 보정하는 접근법이다. 우리는 다섯 가지 제조 사례 연구에서 이러한 기여를 검증한다. 클래스당 참조 이미지 두 장을 사용했을 때 Jev와 Clef-Flash 모두 동일한 참조를 사용한 CLIP의 테스트 정확도를 능가한다(Jev 0.75-1.00 및 Clef-Flash 0.71-0.99 대비 CLIP 0.67-0.84). 또한 텍스트 설명에 대한 에이전트적 개선이 8개 모델-응용 조합 중 6개에서 정확도를 유의미하게 향상시키며, 우리의 보정 접근법이 15개 모델-응용 구성 중 13개에서 Brier 점수를 감소시킴을 보인다. 연구의 재현성을 지원하기 위해 복제 코드를 제공하며 추가 모델을 평가할 수 있는 R "decisions" 패키지를 만들었다. 또한 실무자가 코드 없이 자신의 이미지를 검사할 수 있는 공개 웹 앱도 제공한다.
Automating the classification of camera-obtained microscopic images of White Blood Cells (WBCs) and related cell subtypes has assumed importance since it aids the laborious manual process of review and diagnosis. Several State-Of-The-Art (SOTA) methods developed using Deep Convolutional Neural Networks suffer from the problem of domain shift - severe performance degradation when they are tested on data (target) obtained in a setting different from that of the training (source). The change in the target data might be caused by factors such as differences in camera/microscope types, lenses, lighting-conditions etc. This problem can potentially be solved using Unsupervised Domain Adaptation (UDA) techniques albeit standard algorithms presuppose the existence of a sufficient amount of unlabelled target data which is not always the case with medical images. In this paper, we propose a method for UDA that is devoid of the need for target data. Given a test image from the target data, we obtain its 'closest-clone' from the source data that is used as a proxy in the classifier. We prove the existence of such a clone given that infinite number of data points can be sampled from the source distribution. We propose a method in which a latent-variable generative model based on variational inference is used to simultaneously sample and find the 'closest-clone' from the source distribution through an optimization procedure in the latent space. We demonstrate the efficacy of the proposed method over several SOTA UDA methods for WBC classification on datasets captured using different imaging modalities under multiple settings.
Various methods of measuring unit selectivity have been developed with the aim of better understanding how neural networks work. But the different measures provide divergent estimates of selectivity, and this has led to different conclusions regarding the conditions in which selective object representations are learned and the functional relevance of these representations. In an attempt to better characterize object selectivity, we undertake a comparison of various selectivity measures on a large set of units in AlexNet, including localist selectivity, precision, class-conditional mean activity selectivity (CCMAS), network dissection,the human interpretation of activation maximization (AM) images, and standard signal-detection measures. We find that the different measures provide different estimates of object selectivity, with precision and CCMAS measures providing misleadingly high estimates. Indeed, the most selective units had a poor hit-rate or a high false-alarm rate (or both) in object classification, making them poor object detectors. We fail to find any units that are even remotely as selective as the 'grandmother cell' units reported in recurrent neural networks. In order to generalize these results, we compared selectivity measures on units in VGG-16 and GoogLeNet trained on the ImageNet or Places-365 datasets that have been described as 'object detectors'. Again, we find poor hit-rates and high false-alarm rates for object classification. We conclude that signal-detection measures provide a better assessment of single-unit selectivity compared to common alternative approaches, and that deep convolutional networks of image classification do not learn object detectors in their hidden layers.
Standardized body region labelling of individual images provides data that can improve human and computer use of medical images. A CNN-based classifier was developed to identify body regions in CT and MRI. 17 CT (18 MRI) body regions covering the entire human body were defined for the classification task. Three retrospective databases were built for the AI model training, validation, and testing, with a balanced distribution of studies per body region. The test databases originated from a different healthcare network. Accuracy, recall and precision of the classifier was evaluated for patient age, patient gender, institution, scanner manufacturer, contrast, slice thickness, MRI sequence, and CT kernel. The data included a retrospective cohort of 2,934 anonymized CT cases (training: 1,804 studies, validation: 602 studies, test: 528 studies) and 3,185 anonymized MRI cases (training: 1,911 studies, validation: 636 studies, test: 638 studies). 27 institutions from primary care hospitals, community hospitals and imaging centers contributed to the test datasets. The data included cases of all genders in equal proportions and subjects aged from a few months old to +90 years old. An image-level prediction accuracy of 91.9% (90.2 - 92.1) for CT, and 94.2% (92.0 - 95.6) for MRI was achieved. The classification results were robust across all body regions and confounding factors. Due to limited data, performance results for subjects under 10 years-old could not be reliably evaluated. We show that deep learning models can classify CT and MRI images by body region including lower and upper extremities with high accuracy.
Transferring the ImageNet pre-trained weights to the various remote sensing tasks has produced acceptable results and reduced the need for labeled samples. However, the domain differences between ground imageries and remote sensing images cause the performance of such transfer learning to be limited. Recent research has demonstrated that self-supervised learning methods capture visual features that are more discriminative and transferable than the supervised ImageNet weights. We are motivated by these facts to pre-train the in-domain representations of remote sensing imagery using contrastive self-supervised learning and transfer the learned features to other related remote sensing datasets. Specifically, we used the SimSiam algorithm to pre-train the in-domain knowledge of remote sensing datasets and then transferred the obtained weights to the other scene classification datasets. Thus, we have obtained state-of-the-art results on five land cover classification datasets with varying numbers of classes and spatial resolutions. In addition, By conducting appropriate experiments, including feature pre-training using datasets with different attributes, we have identified the most influential factors that make a dataset a good choice for obtaining in-domain features. We have transferred the features obtained by pre-training SimSiam on remote sensing datasets to various downstream tasks and used them as initial weights for fine-tuning. Moreover, we have linearly evaluated the obtained representations in cases where the number of samples per class is limited. Our experiments have demonstrated that using a higher-resolution dataset during the self-supervised pre-training stage results in learning more discriminative and general representations.
Deep neural networks (DNNs) trained on large-scale datasets have exhibited significant performance in image classification. Many large-scale datasets are collected from websites, however they tend to contain inaccurate labels that are termed as noisy labels. Training on such noisy labeled datasets causes performance degradation because DNNs easily overfit to noisy labels. To overcome this problem, we propose a joint optimization framework of learning DNN parameters and estimating true labels. Our framework can correct labels during training by alternating update of network parameters and labels. We conduct experiments on the noisy CIFAR-10 datasets and the Clothing1M dataset. The results indicate that our approach significantly outperforms other state-of-the-art methods.
Many recent delineation techniques owe much of their increased effectiveness to path classification algorithms that make it possible to distinguish promising paths from others. The downside of this development is that they require annotated training data, which is tedious to produce. In this paper, we propose an Active Learning approach that considerably speeds up the annotation process. Unlike standard ones, it takes advantage of the specificities of the delineation problem. It operates on a graph and can reduce the training set size by up to 80% without compromising the reconstruction quality. We will show that our approach outperforms conventional ones on various biomedical and natural image datasets, thus showing that it is broadly applicable.
Class ambiguity is typical in image classification problems with a large number of classes. When classes are difficult to discriminate, it makes sense to allow k guesses and evaluate classifiers based on the top-k error instead of the standard zero-one loss. We propose top-k multiclass SVM as a direct method to optimize for top-k performance. Our generalization of the well-known multiclass SVM is based on a tight convex upper bound of the top-k error. We propose a fast optimization scheme based on an efficient projection onto the top-k simplex, which is of its own interest. Experiments on five datasets show consistent improvements in top-k accuracy compared to various baselines.
This work proposes a texture descriptor based on fractal theory. The method is based on the Bouligand-Minkowski descriptors. We decompose the original image recursively into 4 equal parts. In each recursion step, we estimate the average and the deviation of the Bouligand-Minkowski descriptors computed over each part. Thus, we extract entropy features from both average and deviation. The proposed descriptors are provided by the concatenation of such measures. The method is tested in a classification experiment under well known datasets, that is, Brodatz and Vistex. The results demonstrate that the proposed technique achieves better results than classical and state-of-the-art texture descriptors, such as Gabor-wavelets and co-occurrence matrix.
From a simplified analysis of adaptive methods, we derive AvaGrad, a new optimizer which outperforms SGD on vision tasks when its adaptability is properly tuned. We observe that the power of our method is partially explained by a decoupling of learning rate and adaptability, greatly simplifying hyperparameter search. In light of this observation, we demonstrate that, against conventional wisdom, Adam can also outperform SGD on vision tasks, as long as the coupling between its learning rate and adaptability is taken into account. In practice, AvaGrad matches the best results, as measured by generalization accuracy, delivered by any existing optimizer (SGD or adaptive) across image classification (CIFAR, ImageNet) and character-level language modelling (Penn Treebank) tasks.
We introduce a model-based asynchronous multi-fidelity method for hyperparameter and neural architecture search that combines the strengths of asynchronous Hyperband and Gaussian process-based Bayesian optimization. At the heart of our method is a probabilistic model that can simultaneously reason across hyperparameters and resource levels, and supports decision-making in the presence of pending evaluations. We demonstrate the effectiveness of our method on a wide range of challenging benchmarks, for tabular data, image classification and language modelling, and report substantial speed-ups over current state-of-the-art methods. Our new methods, along with asynchronous baselines, are implemented in a distributed framework which will be open sourced along with this publication.
Chinese text recognition is more challenging than Latin text due to the large amount of fine-grained Chinese characters and the great imbalance over classes, which causes a serious overfitting problem. We propose to apply Maximum Entropy Regularization to regularize the training process, which is to simply add a negative entropy term to the canonical cross-entropy loss without any additional parameters and modification of a model. We theoretically give the convergence probability distribution and analyze how the regularization influence the learning process. Experiments on Chinese character recognition, Chinese text line recognition and fine-grained image classification achieve consistent improvement, proving that the regularization is beneficial to generalization and robustness of a recognition model.
The continual learning problem has been widely studied in image classification, while rare work has been explored in object detection. Some recent works apply knowledge distillation to constrain the model to retain old knowledge, but this rigid constraint is detrimental for learning new knowledge. In our paper, we propose a new scheme for continual learning of object detection, namely Contrast R-CNN, an approach strikes a balance between retaining the old knowledge and learning the new knowledge. Furthermore, we design a Proposal Contrast to eliminate the ambiguity between old and new instance to make the continual learning more robust. Extensive evaluation on the PASCAL VOC dataset demonstrates the effectiveness of our approach.
AutoML (automated machine learning) has been extensively developed in the past few years for the model-centric approach. As for the data-centric approach, the processes to improve the dataset, such as fixing incorrect labels, adding examples that represent edge cases, and applying data augmentation, are still very artisanal and expensive. Here we develop an automated data-centric tool (AutoDC), similar to the purpose of AutoML, aims to speed up the dataset improvement processes. In our preliminary tests on 3 open source image classification datasets, AutoDC is estimated to reduce roughly 80% of the manual time for data improvement tasks, at the same time, improve the model accuracy by 10-15% with the fixed ML code.
For computer vision, Vision Transformers (ViTs) have become one of the go-to deep net architectures. Despite being inspired by Convolutional Neural Networks (CNNs), ViTs' output remains sensitive to small spatial shifts in the input, i.e., not shift invariant. To address this shortcoming, we introduce novel data-adaptive designs for each of the modules in ViTs, such as tokenization, self-attention, patch merging, and positional encoding. With our proposed modules, we achieve true shift-equivariance on four well-established ViTs, namely, Swin, SwinV2, CvT, and MViTv2. Empirically, we evaluate the proposed adaptive models on image classification and semantic segmentation tasks. These models achieve competitive performance across three different datasets while maintaining 100% shift consistency.
Neural networks are typically sensitive to small input perturbations, leading to unexpected or brittle behaviour. We present RobustNeuralNetworks.jl: a Julia package for neural network models that are constructed to naturally satisfy a set of user-defined robustness constraints. The package is based on the recently proposed Recurrent Equilibrium Network (REN) and Lipschitz-Bounded Deep Network (LBDN) model classes, and is designed to interface directly with Julia's most widely-used machine learning package, Flux.jl. We discuss the theory behind our model parameterization, give an overview of the package, and provide a tutorial demonstrating its use in image classification, reinforcement learning, and nonlinear state-observer design.
Noise injection is a fundamental tool for data augmentation, and yet there is no widely accepted procedure to incorporate it with learning frameworks. This study analyzes the effects of adding or applying different noise models of varying magnitudes to Convolutional Neural Network (CNN) architectures. Noise models that are distributed with different density functions are given common magnitude levels via Structural Similarity (SSIM) metric in order to create an appropriate ground for comparison. The basic results are conforming with the most of the common notions in machine learning, and also introduce some novel heuristics and recommendations on noise injection. The new approaches will provide better understanding on optimal learning procedures for image classification.
We propose KOALA++, a scalable Kalman-based optimization algorithm that explicitly models structured gradient uncertainty in neural network training. Unlike second-order methods, which rely on expensive second order gradient calculation, our method directly estimates the parameter covariance matrix by recursively updating compact gradient covariance products. This design improves upon the original KOALA framework that assumed diagonal covariance by implicitly capturing richer uncertainty structure without storing the full covariance matrix and avoiding large matrix inversions. Across diverse tasks, including image classification and language modeling, KOALA++ achieves accuracy on par or better than state-of-the-art first- and second-order optimizers while maintaining the efficiency of first-order methods.
Knowledge distillation allows transferring knowledge from a pre-trained model to another. However, it suffers from limitations, and constraints related to the two models need to be architecturally similar. Knowledge distillation addresses some of the shortcomings associated with transfer learning by generalizing a complex model to a lighter model. However, some parts of the knowledge may not be distilled by knowledge distillation sufficiently. In this paper, a novel knowledge distillation approach using transfer learning is proposed. The proposed method transfers the entire knowledge of a model to a new smaller one. To accomplish this, unlabeled data are used in an unsupervised manner to transfer the maximum amount of knowledge to the new slimmer model. The proposed method can be beneficial in medical image analysis, where labeled data are typically scarce. The proposed approach is evaluated in the context of classification of images for diagnosing Diabetic Retinopathy on two publicly available datasets, including Messidor and EyePACS. Simulation results demonstrate that the approach is effective in transferring knowledge from a complex model to a lighter one. Furthermore, experimental results illustrate that the performance of different small models is improved significantly using unlabeled data and knowledge distillation.