Binary Convolutional Neural Network Research Articles

BT-CNN: a balanced binary tree architecture for classification of brain tumour using MRI imaging.

Deep learning is a very important technique in clinical diagnosis and therapy in the present world. Convolutional Neural Network (CNN) is a recent development in deep learning that is used in computer vision. Our medical investigation focuses on the identification of brain tumour. To improve the brain tumour classification performance a Balanced binary Tree CNN (BT-CNN) which is framed in a binary tree-like structure is proposed. It has a two distinct modules-the convolution and the depthwise separable convolution group. The usage of convolution group achieves lower time and higher memory, while the opposite is true for the depthwise separable convolution group. This balanced binarty tree inspired CNN balances both the groups to achieve maximum performance in terms of time and space. The proposed model along with state-of-the-art models like CNN-KNN and models proposed by Musallam et al., Saikat et al., and Amin et al. are experimented on public datasets. Before we feed the data into model the images are pre-processed using CLAHE, denoising, cropping, and scaling. The pre-processed dataset is partitioned into training and testing datasets as per 5 fold cross validation. The proposed model is trained and compared its perforarmance with state-of-the-art models like CNN-KNN and models proposed by Musallam et al., Saikat et al., and Amin et al. The proposed model reported average training accuracy of 99.61% compared to other models. The proposed model achieved 96.06% test accuracy where as other models achieved 68.86%, 85.8%, 86.88%, and 90.41% respectively. Further, the proposed model obtained lowest standard deviation on training and test accuracies across all folds, making it invariable to dataset.

A Deep Learning Approach for Surface Crack Classification and Segmentation in Unmanned Aerial Vehicle Assisted Infrastructure Inspections.

Surface crack detection is an integral part of infrastructure health surveys. This work presents a transformative shift towards rapid and reliable data collection capabilities, dramatically reducing the time spent on inspecting infrastructures. Two unmanned aerial vehicles (UAVs) were deployed, enabling the capturing of images simultaneously for efficient coverage of the structure. The suggested drone hardware is especially suitable for the inspection of infrastructure with confined spaces that UAVs with a broader footprint are incapable of accessing due to a lack of safe access or positioning data. The collected image data were analyzed using a binary classification convolutional neural network (CNN), effectively filtering out images containing cracks. A comparison of state-of-the-art CNN architectures against a novel CNN layout "CrackClassCNN" was investigated to obtain the optimal layout for classification. A Segment Anything Model (SAM) was employed to segment defect areas, and its performance was benchmarked against manually annotated images. The suggested "CrackClassCNN" achieved an accuracy rate of 95.02%, and the SAM segmentation process yielded a mean Intersection over Union (IoU) score of 0.778 and an F1 score of 0.735. It was concluded that the selected UAV platform, the communication network, and the suggested processing techniques were highly effective in surface crack detection.

Fully Pipelined FPGA Acceleration of Binary Convolutional Neural Networks with Neural Architecture Search

In this paper, we present a fully pipelined and semi-parallel channel convolutional neural network hardware accelerator structure. This structure can trade off the compute time and the hardware utilization, allowing the accelerator to be layer pipelined without the need for fully parallelizing the input and output channels. A parallel strategy is applied to reduce the time gap in transferring the output results between different layers. The parallelism can be decided based on the hardware resources on the target FPGA. We use this structure to implement a binary ResNet18 based on the neural architecture search strategy, which can increase the accuracy of manually designed binary convolutional neural networks. Our optimized binary ResNet18 can achieve a Top-1 accuracy of 60.5% on the ImageNet dataset. We deploy this ResNet18 hardware implementation on an Alphadata 9H7 FPGA, connected with an OpenCAPI interface, to demonstrate the hardware capabilities. Depending on the amount of parallelism used, the latency can range from 1.12 to 6.33[Formula: see text]ms, with a corresponding throughput of 4.56 to 0.71 TOPS for different hardware utilization, with a 200[Formula: see text]MHz clock frequency. Our best latency is [Formula: see text] lower and our best throughput is [Formula: see text] higher compared to the best previous works. The code for our implementation is open-source and publicly available on GitHub at https://github.com/MFJI/NASBRESNET .

Interfacial magnetic spin Hall effect in van der Waals Fe3GeTe2/MoTe2 heterostructure

The spin Hall effect (SHE) allows efficient generation of spin polarization or spin current through charge current and plays a crucial role in the development of spintronics. While SHE typically occurs in non-magnetic materials and is time-reversal even, exploring time-reversal-odd (T-odd) SHE, which couples SHE to magnetization in ferromagnetic materials, offers a new charge-spin conversion mechanism with new functionalities. Here, we report the observation of giant T-odd SHE in Fe3GeTe2/MoTe2 van der Waals heterostructure, representing a previously unidentified interfacial magnetic spin Hall effect (interfacial-MSHE). Through rigorous symmetry analysis and theoretical calculations, we attribute the interfacial-MSHE to a symmetry-breaking induced spin current dipole at the vdW interface. Furthermore, we show that this linear effect can be used for implementing multiply-accumulate operations and binary convolutional neural networks with cascaded multi-terminal devices. Our findings uncover an interfacial T-odd charge-spin conversion mechanism with promising potential for energy-efficient in-memory computing.

Detection and quantitative analysis of patient-ventilator interactions in ventilated infants by deep learning networks.

The study of patient-ventilator interactions (PVI) in mechanically ventilated neonates is limited by the lack of unified PVI definitions and tools to perform large scale analyses. An observational study was conducted in 23 babies randomly selected from 170 neonates who were ventilated with SIPPV-VG, SIMV-VG or PSV-VG mode for at least 12 h. 500 breaths were randomly selected and manually annotated from each recording to train convolutional neural network (CNN) models for PVI classification. The average asynchrony index (AI) over all recordings was 52.5%. The most frequently occurring PVIs included expiratory work (median: 28.4%, interquartile range: 23.2-40.2%), late cycling (7.6%, 2.8-10.2%), failed triggering (4.6%, 1.2-6.2%) and late triggering (4.4%, 2.8-7.4%). Approximately 25% of breaths with a PVI had two or more PVIs occurring simultaneously. Binary CNN classifiers were developed for PVIs affecting ≥1% of all breaths (n = 7) and they achieved F1 scores of >0.9 on the test set except for early triggering where it was 0.809. PVIs occur frequently in neonates undergoing conventional mechanical ventilation with a significant proportion of breaths containing multiple PVIs. We have developed computational models for seven different PVIs to facilitate automated detection and further evaluation of their clinical significance in neonates. The study of patient-ventilator interactions (PVI) in mechanically ventilated neonates is limited by the lack of unified PVI definitions and tools to perform large scale analyses. By adapting a recent taxonomy of PVI definitions in adults, we have manually annotated neonatal ventilator waveforms to determine prevalence and co-occurrence of neonatal PVIs. We have also developed binary deep learning classifiers for common PVIs to facilitate their automatic detection and quantification.

One-Dimensional Binary Convolutional Neural Network Accelerator Design for Bearing Fault Diagnosis

One-Dimensional Binary Convolutional Neural Network Accelerator Design for Bearing Fault Diagnosis

A hybrid LBP-CNN with YOLO-v5-based fire and smoke detection model in various environmental conditions for environmental sustainability in smart city.

Smart, secure, and environmentally friendly smart cities are all the rage in urban planning. Several technologies, including the Internet of Things (IoT) and edge computing, are used to develop smart cities. Early and accurate fire detection in a Smart city is always desirable and motivates the research community to create a more efficient model. Deep learning models are widely used for fire detection in existing research, but they encounter several issues in typical climate environments, such as foggy and normal. The proposed model lends itself to IoT applications for authentic fire surveillance because of its minimal configuration load. A hybrid Local Binary Pattern Convolutional Neural Network (LBP-CNN) and YOLO-V5 model-based fire detection model for smart cities in the foggy scenario is presented in this research. Additionally, we recommend a two-part technique for extracting features to be applied to YOLO throughout this article. Using a transfer learning technique, the first portion of the proposed approach for extracting features retrieves standard features. The section part is for retrieval of additional valuable information related to the current activity using the LBP (Local Binary Pattern) protective layer and classifications layers. This research utilizes an online Kaggle fire and smoke dataset with 13950 normal and foggy images. The proposed hybrid model is premised on a two-cascaded YOLO model. In the initial cascade, smoke and fire are detected in the normal surrounding region, and the second cascade fire is detected with density in a foggy environment. In experimental analysis, the proposed model achieved a fire and smoke detection precision rate of 96.25% for a normal setting, 93.2% for a foggy environment, and a combined detection average precision rate of 94.59%. The proposed hybrid system outperformed existing models in terms of better precision and density detection for fire and smoke.

P469 Deep learning and minimally invasive endoscopy – panendoscopic detection of pleomorphic lesions

Abstract Background Capsule endoscopy is a minimally invasive exam capable of performing a panendoscopic evaluation of the gastrointestinal tract. Nevertheless, capsule endoscopy is a time-consuming exam, revealing suboptimal diagnostic yield when considering the upper gastrointestinal tract. Convolutional neural network are models based on the human visual cortex architecture, suitable for image analysis. However, there is still an absence of studies about their role in capsule panendoscopy. Methods Our group developed a deep Learning model for panendoscopic automatic detection of pleomorphic lesions (namely vascular lesions, protuberant lesions, hematic residues, ulcers and erosions). In order to develop the convolutional neural network, 355 110 images (6 977 oesophageal, 12 918 gastric, 258 443 enteric, 76 772 colonic) from eight different capsule endoscopy and colon capsule endoscopy devices were divided in a training and validation dataset in a patient split design. The model’s output was compared to three CE experts’ classification. The model performance was evaluated by its sensitivity, specificity, accuracy, positive predictive value, negative predictive value, and area under the precision-recall curve. Results The panendoscopic convolutional neural network was composed by organ-specific neural networks. The binary oesophagus convolutional neural network had a diagnostic accuracy for pleomorphic lesions of 83.6%. The binary gastric network identified pleomorphic lesions with a 96.6% accuracy. The undenary small bowel convolutional neural network distinguished pleomorphic lesions with different potential with 97.6% accuracy. The trinary colonic model (detection and differentiation of normal mucosa, pleomorphic lesions and hematic residues) had 94.9% global accuracy. Conclusion Our group developed the first deep learning model for panendoscopic automatic detection of pleomorphic lesions in both small bowel and colon capsule endoscopy devices from multiple brands, solving a critical interoperability technological challenge. Deep learning-based tools may change the landscape of minimally invasive capsule panendoscopy.

Multibit, Lead‐Free Cs2SnI6 Resistive Random Access Memory with Self‐Compliance for Improved Accuracy in Binary Neural Network Application

AbstractIn the realm of neuromorphic computing, integrating Binary Neural Networks (BNN) with non‐volatile memory based on emerging materials can be a promising avenue for introducing novel functionalities. This study underscores the viability of lead‐free, air‐stable Cs2SnI6 (CSI) based resistive random access memory (RRAM) devices as synaptic weights in neuromorphic architectures, specifically for BNNs applications. Herein, hydrothermally synthesized CSI perovskites are explored as a resistive layer in RRAM devices either on the rigid or flexible substrate, highlighting reproducible multibit switching with self‐compliance, low‐ resistance‐state (LRS) variations, a decent On/Off ratio(or retention) of ≈103(or 104 s), and endurance exceeding 300 cycles. Moreover, a comprehensive evaluation with the 32 × 32 × 3 RGB CIFAR‐10 dataset reveals that binary convolutional neural networks (BCNN) trained solely on binary weight values can achieve competitive rates of accuracy comparable to those of their analog weight counterparts. These findings highlight the dominance of the LRS for CSI RRAM with self‐compliance in a weighted configuration and minimal influence of the high resistance state despite substantial fluctuations for flexible CSI RRAM under varying bending radii. With its unique electrical switching capabilities, the CSI RRAM is highly anticipated to emerge as a promising candidate for embedded AI systems, especially in IoT devices and wearables.

Optimal performance of Binary Relevance CNN in targeted multi-label text classification

In the context of multi-label text classification (MLTC), Binary Relevance (BR) stands out as one of the most intuitive and frequently employed methodologies. It tackles the MLTC task by breaking it down into multiple binary classification problems. However, BR has faced conceptual criticism due to its omission of label dependency information. To address this limitation, numerous studies have concentrated their efforts on enhancing the incorporation of label dependencies and document features. This resulted in substantial improvements in the performance of MLTC models. While the question of whether models incorporating label dependency information consistently outperform BR models remains unanswered, the prevailing opinion suggests their superiority. In this paper, we present evidence that challenges this widely held belief. Our numerical results across various text datasets demonstrate that an optimized binary relevance convolutional neural network (BR-CNN) can outperform advanced multi-label learning models explicitly designed to leverage label dependency information as well as advanced Binary Relevance (BR) models. Our result underscores the competitiveness of a BR-CNN approach for MLTC and emphasizes the versatility of the BR model family as a customizable option. More fundamentally, our findings contribute to the ongoing discourse surrounding label dependency and provide valuable insights into the efficacy of the binary relevance approach.

Machine learning studies for liquid crystal texture recognition

ABSTRACT Machine learning techniques such as local binary pattern algorithm and convolutional neural network were applied to study textures of liquid crystal compound (S)-4’-(1-methyloctyloxycarbonyl) biphenyl-4-yl 4-[7-(2,2,3,3,4,4,4-heptafluorobutoxy) heptyl-1-oxy]-benzoate (3F7HPhH7). This compound exhibits in its polymorphism several smectic phases (smectic A, ferroelectric smectic C and antiferroelectric smectic C) and two glass states of antiferroelectric smectic C phase, which textures are difficult to distinguish. Proof-of-concept evidence is provided, demonstrating the ability of machine learning algorithms to identify transition temperatures and the respective phases involved. The article describes the procedure for preparing a dataset of textures obtained from polarised microscopy for classification using convolutional neural networks, especially in cases where the image dataset is unbalanced. It utilises the feature of their self-similarity and image augmentation, particularly the cropping procedure. A Kolmogorov–Smirnov test was conducted to check if selected images do not carry the same information.

Time-Domain Measurement Data Accumulation for Slow Moving Point Target Detection in Heavily Cluttered Environments Using CNN

In modern radars, the target detection probability is increased by lowering the detection threshold via signal processing to detect a point target with a small radar cross-section value. However, a lower threshold increases the number of false targets. In the conventional tracking method, which uses a general tracking filter, the measurement data between scans should be compared. Therefore, for a large amount of acquired measurement data, the computational complexity can be reduced by accumulating the acquired measurement data over time, recognizing the target movement as a pattern, and training a convolutional neural network (CNN) model. Here, we propose a method to create a desired target scenario by transfer learning and estimate the target position using the activation map of a binary detector CNN model. The model can detect a target using the actual acquired radar data, and the processing time remains constant, regardless of the number of false alarms.

Voice spoofing detection for multiclass attack classification using deep learning

Voice biometric authentication is increasingly gaining adoption in organisations with high-volume identity verifications and for providing access to physical and other virtual spaces. In this form of authentication, the user’s identity is verified with their voice. However, these systems are susceptible to voice spoofing attacks as malicious actors employ different types of attacks such as speech synthesis, voice conversion or imitations, and recorded replays to spoof the Automatic Speaker Verification (ASV) system or for spam communications. In this work, we provide a voice spoofing countermeasure as a binary classification problem, that classifies real and fake audio, and also as a multiclass classification problem to detect voice conversion, synthesis and replay attacks. We investigated numerous audio features and examined each feature capability alongside state-of-the-art deep learning algorithms including convolutional neural networks (CNN), WaveNet, and recurrent neural network variants — Gated Recurrent Unit (GRU) and Long Short-Term Memory (LSTM) models. Using a large dataset of 419,426 audio files for experiments, we evaluated the deep learning models for their effectiveness against voice spoofing attacks. The binary class CNN achieved a false positive rate (FPR) of 0.0216, while the multiclass solutions using CNN, WaveNet, LSTMs and GRUs achieved an FPR of 0.003, 0.0260, 0.0302 and 0.0358 respectively. We extended the evaluation of the models by including the real-time classification using microphone voice audio and user-uploaded audio to demonstrate the practical implications and deployability.

Invariant Attribute-Driven Binary Bi-Branch Classification of Hyperspectral and LiDAR Images

The fusion of hyperspectral and LiDAR images plays a crucial role in remote sensing by capturing spatial relationships and modeling semantic information for accurate classification and recognition. However, existing methods, such as Graph Convolutional Networks (GCNs), face challenges in constructing effective graph structures due to variations in local semantic information and limited receptiveness to large-scale contextual structures. To overcome these limitations, we propose an Invariant Attribute-driven Binary Bi-branch Classification (IABC) method, which is a unified network that combines a binary Convolutional Neural Network (CNN) and a GCN with invariant attributes. Our approach utilizes a joint detection framework that can simultaneously learn features from small-scale regular regions and large-scale irregular regions, resulting in an enhanced structural representation of HSI and LiDAR images in the spectral–spatial domain. This approach not only improves the accuracy of classification and recognition but also reduces storage requirements and enables real-time decision making, which is crucial for effectively processing large-scale remote sensing data. Extensive experiments demonstrate the superior performance of our proposed method in hyperspectral image analysis tasks. The combination of CNNs and GCNs allows for the accurate modeling of spatial relationships and effective construction of graph structures. Furthermore, the integration of binary quantization enhances computational efficiency, enabling the real-time processing of large-scale data. Therefore, our approach presents a promising opportunity for advancing remote sensing applications using deep learning techniques.

Photonic Binary Convolutional Neural Network Based on Microring Resonator Array

We propose the photonic architecture based on microring resonator (MRR) arrays for binary convolutional neural networks (BCNNs) accelerated computing. The MRR crossbar unit is used for computing weight {−1, 1} and the single MRR is for input {0, 1}. The computing parallelism is improved through wavelength division multiplexing. The photonic BCNN achieves 97.29% classification accuracy on the MNIST test set which is only 1.94% lower than the accuracy of the 32-bit neural network, and saves <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$32\times $ </tex-math></inline-formula> at memory usage. We analyze effects of input and weight encoding errors on the photonic BCNN. When the input or weight error rate is less than 0.01%, the test accuracy remains unchanged. We evaluate the performance of the photonic BCNN architecture considering optical loss, inter-channel crosstalk, operation frequency and device power consumption. The energy efficiency of the designed photonic BCNN architecture is 1.72 pJ/MAC, which is <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$4.80\times $ </tex-math></inline-formula> and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$61.32\times $ </tex-math></inline-formula> better than the 8-bit and 16-bit architecture respectively. The photonic BCNN is promising to be used for edge computing.

An ADC-Less RRAM-Based Computing-in-Memory Macro With Binary CNN for Efficient Edge AI

Resistive random-access memory (RRAM) based non-volatile computing-in-memory (nvCIM) has been regarded as a promising solution to enable efficient data-intensive artificial intelligence (AI) applications on resource-limited edge systems. However, existing weighted-current summation-based nvCIM suffers from device non-idealities and significant time, storage, and energy overheads when processing high-precision analog signals. To address these issues, we propose a 3T2R digital nvCIM macro for a fully hardware-implemented binary convolutional neural network (HBCNN), focusing on accelerating edge AI applications at low weight precision. By quantizing the voltage-division results of RRAMs through inverters, the 3T2R macro provides a stable rail-to-rail output without analog-to-digital converters or sensing amplifiers. Moreover, both batch normalization and sign activation are integrated on-chip. The hybrid simulation results show that the proposed 3T2R digital macro achieves an 86.2% (95.6%) accuracy on the CIFAR-10 (MNIST) dataset, corresponding to a 4.7% (1.9%) accuracy loss compared to the software baselines, which also feature a peak energy efficiency of 51.3 TOPS/W and a minimum latency of 8 ns, realizing an energyefficient, low-latency, and robust AI processor.

Artificially-generated consolidations and balanced augmentation increase performance of U-net for lung parenchyma segmentation on MR images.

To improve automated lung segmentation on 2D lung MR images using balanced augmentation and artificially-generated consolidations for training of a convolutional neural network (CNN). From 233 healthy volunteers and 100 patients, 1891 coronal MR images were acquired. Of these, 1666 images without consolidations were used to build a binary semantic CNN for lung segmentation and 225 images (187 without consolidations, 38 with consolidations) were used for testing. To increase CNN performance of segmenting lung parenchyma with consolidations, balanced augmentation was performed and artificially-generated consolidations were added to all training images. The proposed CNN (CNNBal/Cons) was compared to two other CNNs: CNNUnbal/NoCons-without balanced augmentation and artificially-generated consolidations and CNNBal/NoCons-with balanced augmentation but without artificially-generated consolidations. Segmentation results were assessed using Sørensen-Dice coefficient (SDC) and Hausdorff distance coefficient. Regarding the 187 MR test images without consolidations, the mean SDC of CNNUnbal/NoCons (92.1 ± 6% (mean ± standard deviation)) was significantly lower compared to CNNBal/NoCons (94.0 ± 5.3%, P = 0.0013) and CNNBal/Cons (94.3 ± 4.1%, P = 0.0001). No significant difference was found between SDC of CNNBal/Cons and CNNBal/NoCons (P = 0.54). For the 38 MR test images with consolidations, SDC of CNNUnbal/NoCons (89.0 ± 7.1%) was not significantly different compared to CNNBal/NoCons (90.2 ± 9.4%, P = 0.53). SDC of CNNBal/Cons (94.3 ± 3.7%) was significantly higher compared to CNNBal/NoCons (P = 0.0146) and CNNUnbal/NoCons (P = 0.001). Expanding training datasets via balanced augmentation and artificially-generated consolidations improved the accuracy of CNNBal/Cons, especially in datasets with parenchymal consolidations. This is an important step towards a robust automated postprocessing of lung MRI datasets in clinical routine.

Can Machine Learning Be Better than Biased Readers?

Background: Training machine learning (ML) models in medical imaging requires large amounts of labeled data. To minimize labeling workload, it is common to divide training data among multiple readers for separate annotation without consensus and then combine the labeled data for training a ML model. This can lead to a biased training dataset and poor ML algorithm prediction performance. The purpose of this study is to determine if ML algorithms can overcome biases caused by multiple readers' labeling without consensus. Methods: This study used a publicly available chest X-ray dataset of pediatric pneumonia. As an analogy to a practical dataset without labeling consensus among multiple readers, random and systematic errors were artificially added to the dataset to generate biased data for a binary-class classification task. The Resnet18-based convolutional neural network (CNN) was used as a baseline model. A Resnet18 model with a regularization term added as a loss function was utilized to examine for improvement in the baseline model. Results: The effects of false positive labels, false negative labels, and random errors (5-25%) resulted in a loss of AUC (0-14%) when training a binary CNN classifier. The model with a regularized loss function improved the AUC (75-84%) over that of the baseline model (65-79%). Conclusion: This study indicated that it is possible for ML algorithms to overcome individual readers' biases when consensus is not available. It is recommended to use regularized loss functions when allocating annotation tasks to multiple readers as they are easy to implement and effective in mitigating biased labels.

Local Binary Convolutional Neural Networks' Long Short-Term Memory Model for Human Embryos' Anomaly Detection

Accurate selection of embryos with the maximum implementation condition is a necessary step to increase the effectiveness of fertility treatment in in vitro fertilization (IVF). The deep learning algorithms presented high potential for monitoring and visualizing embryo features such as cell numbers and their morphological development in time series manner. Due to the ability of the computer vision and deep learning algorithms, this paper aimed to present a novel deep learning approach to distinguish simultaneous abnormality of embryos in time-lapse systems for detecting live and non-live births in IVF. The approach is composed of local binary convolutional neural network (LBCNN) and long short-term memory (LSTM). The LBCNN improved accuracy of classification by employing deep and local feature sets with lower number of learnable parameters in comparison with a standard convolutional layer. Moreover, LSTM network is employed to analyze temporal information of time-lapse embryos. The results indicate that the proposed approach achieves significant results in ROC analysis (0.98) in 5 days of monitoring compared to state-of-the-art models. In addition, the approach showed compatible results in early diagnosis of abnormality detection (72 hours) with 82.8% accuracy of classification compared to the pretrained well-known convolutional neural network (CNN) models and baseline CNN.

A Lightweight Binarized Convolutional Neural Network Model for Small Memory and Low-Cost Mobile Devices

In recent years, the high cost of implementing deep neural networks due to their large model size and parameter complexity has made it a challenging problem to design lightweight models that reduce application costs. The existing binarized neural networks suffer from both the large memory occupancy and the big number of trainable params they use. We propose a lightweight binarized convolutional neural network (CBCNN) model to address the multiclass classification/identification problem. We use both binary weights and activation. We show experimentally that a model using only 0.59 M trainable params is sufficient to reach about 92.94% accuracy on the GTSRB dataset, and it has similar performances compared to other methods on MNIST and Fashion-MNIST datasets. Accordingly, most arithmetic operations with bitwise operations are simplified, thus both used memory size and memory accesses are reduced by 32 times. Moreover, the color information was removed, which also reduced drastically the training time. All these together allow our architecture to run effectively and in real time on simple CPUs (rather than GPUs). Through the results we obtained, we show that despite simplifications and color information removal, our network achieves similar performances compared to classical CNNs with lower costs in both in training and embedded deployment.