Claim Missing Document
Check
Articles

Found 2 Documents
Search

Object detection for waste management: a comparative review of models, challenges, and future directions Owen Tamin; Ervin Gubin Moung; Ali Farzamnia
Bulletin of Electrical Engineering and Informatics Vol 15, No 3: June 2026
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eei.v15i3.10696

Abstract

Despite growing interest in automated waste detection, existing surveys either focus on a narrow set of models or lack systematic comparisons across object detection paradigms. This review addresses that gap by examining recent advances in deep learning for waste management, spanning two-stage detectors (Faster region-based convolutional neural network (Faster R-CNN) and Mask region-based convolutional neural network (Mask R-CNN)), single-shot frameworks (you only look once version 1 (YOLO)v1 to YOLOv11), and emerging Transformer-based models (ViT-WM and AL-DETR). Faster R-CNN achieved category-level accuracy of 91.68% and overall accuracy of 89.68%, while Mask R-CNN reported AP values between 26.2% and 34.5% across varied datasets. YOLO models demonstrated strong real-time capability, with YOLOv5 reaching a mAP@0.5 of 92.96% and YOLOv8 achieving 97.63% accuracy with precision and recall above 93%. Transformer-based approaches are especially promising: ViT-WM achieved 98.17% accuracy, the highest among reviewed models, and AL-DETR reported a mAP of 58.9% while integrating active learning (AL) strategies to reduce reliance on extensive labeled data. These results emphasize YOLO’s efficiency for real-time waste sorting and the potential of Transformer architectures for handling complex, cluttered environments. Remaining challenges include dataset variability, computational demand, and limited standardized benchmarks. Future research should prioritize developing comprehensive datasets, optimizing Transformers for real-time use, and leveraging AL to enhance generalizability with reduced annotation effort.
Trade-Offs in Statistical and Machine-Learning-Based Imputation for Algal Time-Series: A Multi-Criteria Evaluation on Optical Density Data Without Ground Truth Erickson James; Ervin Gubin Moung; Mohd Shahazureen Ikwan Bin Abdul Rahman; Valentino Liaw; Lucky Goh Poh Wah; Ahmad Asnawi Bin Mus; Ali Farzamnia
Journal of Applied Data Sciences Vol 7, No 4: December 2026
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v7i4.1539

Abstract

Missing values are pervasive in algal cultivation monitoring, where instrument failure and transmission loss remove whole runs of consecutive observations, and the repair method chosen silently shapes every downstream analysis. This study asks which repair method should be used, and on what evidence, when no complete reference series exists. Twelve methods are compared on optical density and ten covariates recorded across 30 open-pond production cycles at five geographically distinct sites: four interpolation schemes, a propagation scheme, two nearest-neighbour schemes, three chained-equation variants, a random-forest ensemble and a mechanistic growth-curve fit. The principal contribution is a validation framework organised in two clearly separate tiers. The first tier needs no ground truth and applies to the naturally missing values, comparing central tendency, dispersion, distributional shape, autocorrelation structure and coverage between observed and repaired series. The second tier withholds values that were in fact recorded and rebuilds them, so reconstruction error can be measured directly; it is replicated over ten random seeds, three masking proportions and four contiguous outage lengths, giving 25,200 paired evaluations in which every method sees an identical pattern of withhold cells. Key findings found that no method dominates. Ensemble and chained-equation learners rebuild individual values most accurately, with a normalised error of 0.1927 against 0.2645 for linear interpolation, but they compress dispersion and distort distributional shape, whereas shape-preserving interpolation and propagation protect those properties and rebuild less accurately. Gap structure matters: interpolation loses accuracy when gaps are contiguous rather than scattered, while multivariate learners gain. Regularising the chained equations removes a convergence failure that left one pond unrepaired. An aggregated decision analysis turns these trade-offs into a weight-dependent selection rule, giving practitioners an auditable basis for matching method to analytical objective instead of a single universal default.