-
Lightweight Deep Models for Dermatological Disease Detection: A Study on Instance Selection and Channel Optimization
Authors:
Ian Mateos Gonzalez,
Estefani Jaramilla Nava,
Abraham Sánchez Morales,
Jesús García-Ramírez,
Ricardo Ramos-Aguilar
Abstract:
The identification of dermatological disease is an important problem in Mexico according with different studies. Several works in literature use the datasets of different repositories without applying a study of the data behavior, especially in medical images domain. In this work, we propose a methodology to preprocess dermaMNIST dataset in order to improve its quality for the classification stage…
▽ More
The identification of dermatological disease is an important problem in Mexico according with different studies. Several works in literature use the datasets of different repositories without applying a study of the data behavior, especially in medical images domain. In this work, we propose a methodology to preprocess dermaMNIST dataset in order to improve its quality for the classification stage, where we use lightweight convolutional neural networks. In our results, we reduce the number of instances for the neural network training obtaining a similar performance of models as ResNet.
△ Less
Submitted 1 April, 2025;
originally announced April 2025.
-
Enhancing Predictive Maintenance in Mining Mobile Machinery through a TinyML-enabled Hierarchical Inference Network
Authors:
Raúl de la Fuente,
Luciano Radrigan,
Anibal S Morales
Abstract:
Mining machinery operating in variable environments faces high wear and unpredictable stress, challenging Predictive Maintenance (PdM). This paper introduces the Edge Sensor Network for Predictive Maintenance (ESN-PdM), a hierarchical inference framework across edge devices, gateways, and cloud services for real-time condition monitoring. The system dynamically adjusts inference locations--on-devi…
▽ More
Mining machinery operating in variable environments faces high wear and unpredictable stress, challenging Predictive Maintenance (PdM). This paper introduces the Edge Sensor Network for Predictive Maintenance (ESN-PdM), a hierarchical inference framework across edge devices, gateways, and cloud services for real-time condition monitoring. The system dynamically adjusts inference locations--on-device, on-gateway, or on-cloud--based on trade-offs among accuracy, latency, and battery life, leveraging Tiny Machine Learning (TinyML) techniques for model optimization on resource-constrained devices. Performance evaluations showed that on-sensor and on-gateway inference modes achieved over 90\% classification accuracy, while cloud-based inference reached 99\%. On-sensor inference reduced power consumption by approximately 44\%, enabling up to 104 hours of operation. Latency was lowest for on-device inference (3.33 ms), increasing when offloading to the gateway (146.67 ms) or cloud (641.71 ms). The ESN-PdM framework provides a scalable, adaptive solution for reliable anomaly detection and PdM, crucial for maintaining machinery uptime in remote environments. By balancing accuracy, latency, and energy consumption, this approach advances PdM frameworks for industrial applications.
△ Less
Submitted 16 November, 2024; v1 submitted 11 November, 2024;
originally announced November 2024.
-
Challenging mitosis detection algorithms: Global labels allow centroid localization
Authors:
Claudio Fernandez-Martín,
Umay Kiraz,
Julio Silva-Rodríguez,
Sandra Morales,
Emiel Janssen,
Valery Naranjo
Abstract:
Mitotic activity is a crucial proliferation biomarker for the diagnosis and prognosis of different types of cancers. Nevertheless, mitosis counting is a cumbersome process for pathologists, prone to low reproducibility, due to the large size of augmented biopsy slides, the low density of mitotic cells, and pattern heterogeneity. To improve reproducibility, deep learning methods have been proposed…
▽ More
Mitotic activity is a crucial proliferation biomarker for the diagnosis and prognosis of different types of cancers. Nevertheless, mitosis counting is a cumbersome process for pathologists, prone to low reproducibility, due to the large size of augmented biopsy slides, the low density of mitotic cells, and pattern heterogeneity. To improve reproducibility, deep learning methods have been proposed in the last years using convolutional neural networks. However, these methods have been hindered by the process of data labelling, which usually solely consist of the mitosis centroids. Therefore, current literature proposes complex algorithms with multiple stages to refine the labels at pixel level, and to reduce the number of false positives. In this work, we propose to avoid complex scenarios, and we perform the localization task in a weakly supervised manner, using only image-level labels on patches. The results obtained on the publicly available TUPAC16 dataset are competitive with state-of-the-art methods, using only one training phase. Our method achieves an F1-score of 0.729 and challenges the efficiency of previous methods, which required multiple stages and strong mitosis location information.
△ Less
Submitted 30 November, 2022;
originally announced November 2022.
-
High precision indoor positioning by means of LiDAR
Authors:
Eduardo Sánchez Morales,
Michael Botsch,
Bertold Huber,
Andrés García Higuera
Abstract:
The trend towards autonomous driving and the continuous research in the automotive area, like Advanced Driver Assistance Systems (ADAS), requires an accurate localization under all circumstances. An accurate estimation of the vehicle state is a basic requirement for any trajectory-planning algorithm. Still, even when the introduction of the GPS L5 band promises lane-accuracy, coverage limitations…
▽ More
The trend towards autonomous driving and the continuous research in the automotive area, like Advanced Driver Assistance Systems (ADAS), requires an accurate localization under all circumstances. An accurate estimation of the vehicle state is a basic requirement for any trajectory-planning algorithm. Still, even when the introduction of the GPS L5 band promises lane-accuracy, coverage limitations in roofed areas still have to be addressed. In this work, a method for high precision indoor positioning using a LiDAR is presented. The method is based on the combination of motion models with LiDAR measurements, and uses infrastructural elements as positioning references. This allows to estimate the orientation, velocity over ground and position of a vehicle in a Local Tangent Plane (LTP) reference frame. When the outputs of the proposed method are compared to those of an Automotive Dynamic Motion Analyzer (ADMA), mean errors of 1 degree, 0.1 m/s and of 4.7 cm respectively are obtained. The method can be implemented by using a LiDAR sensor as a stand-alone unit. A median runtime of 40.77 us on an Intel i7-6820HQ CPU signals the possibility of real-time processing.
△ Less
Submitted 14 May, 2020;
originally announced May 2020.
-
High Precision Indoor Navigation for Autonomous Vehicles
Authors:
Eduardo Sánchez Morales,
Michael Botsch,
Bertold Huber,
Andrés García Higuera
Abstract:
Autonomous driving is an important trend of the automotive industry. The continuous research towards this goal requires a precise reference vehicle state estimation under all circumstances in order to develop and test autonomous vehicle functions. However, even when lane-accurate positioning is expected from oncoming technologies, like the L5 GPS band, the question of accurate positioning in roofe…
▽ More
Autonomous driving is an important trend of the automotive industry. The continuous research towards this goal requires a precise reference vehicle state estimation under all circumstances in order to develop and test autonomous vehicle functions. However, even when lane-accurate positioning is expected from oncoming technologies, like the L5 GPS band, the question of accurate positioning in roofed areas, e.\,g., tunnels or park houses, still has to be addressed.
In this paper, a novel procedure for a reference vehicle state estimation is presented. The procedure includes three main components. First, a robust standstill detection based purely on signals from an Inertial Measurement Unit. Second, a vehicle state estimation by means of statistical filtering. Third, a high accuracy LiDAR-based positioning method that delivers velocity, position and orientation correction data with a mean error of 0.1 m/s, 4.7 cm and 1$^\circ$ respectively. Runtime tests on a CPU indicates the possibility of real-time implementation.
△ Less
Submitted 14 May, 2020;
originally announced May 2020.
-
Accuracy Characterization of the Vehicle State Estimation from Aerial Imagery
Authors:
Eduardo Sánchez Morales,
Friedrich Kruber,
Michael Botsch,
Bertold Huber,
Andrés García Higuera
Abstract:
Due to their capability of acquiring aerial imagery, camera-equipped Unmanned Aerial Vehicles (UAVs) are very cost-effective tools for acquiring traffic information. However, not enough attention has been given to the validation of the accuracy of these systems. In this paper, an analysis of the most significant sources of error is done. This includes three key components. First, a vehicle state e…
▽ More
Due to their capability of acquiring aerial imagery, camera-equipped Unmanned Aerial Vehicles (UAVs) are very cost-effective tools for acquiring traffic information. However, not enough attention has been given to the validation of the accuracy of these systems. In this paper, an analysis of the most significant sources of error is done. This includes three key components. First, a vehicle state estimation by means of statistical filtering. Second, a quantification of the most significant sources of error. Third, a benchmark of the estimated state compared with state-of-the-art reference sensors. This work presents ways to minimize the errors of the most relevant sources. With these error reductions, camera-equipped UAVs are very attractive tools for traffic data acquisition. The test data and the source code are made publicly available.
△ Less
Submitted 13 May, 2020;
originally announced May 2020.
-
Unsupervised and Supervised Learning with the Random Forest Algorithm for Traffic Scenario Clustering and Classification
Authors:
Friedrich Kruber,
Jonas Wurst,
Eduardo Sánchez Morales,
Samarjit Chakraborty,
Michael Botsch
Abstract:
The goal of this paper is to provide a method, which is able to find categories of traffic scenarios automatically. The architecture consists of three main components: A microscopic traffic simulation, a clustering technique and a classification technique for the operational phase. The developed simulation tool models each vehicle separately, while maintaining the dependencies between each other.…
▽ More
The goal of this paper is to provide a method, which is able to find categories of traffic scenarios automatically. The architecture consists of three main components: A microscopic traffic simulation, a clustering technique and a classification technique for the operational phase. The developed simulation tool models each vehicle separately, while maintaining the dependencies between each other. The clustering approach consists of a modified unsupervised Random Forest algorithm to find a data adaptive similarity measure between all scenarios. As part of this, the path proximity, a novel technique to determine a similarity based on the Random Forest algorithm is presented. In the second part of the clustering, the similarities are used to define a set of clusters. In the third part, a Random Forest classifier is trained using the defined clusters for the operational phase. A thresholding technique is described to ensure a certain confidence level for the class assignment. The method is applied for highway scenarios. The results show that the proposed method is an excellent approach to automatically categorize traffic scenarios, which is particularly relevant for testing autonomous vehicle functionality.
△ Less
Submitted 5 April, 2020;
originally announced April 2020.
