Science AI Journal
Computer Science · Research topic

Open research questions in Image Retrieval and Classification Techniques

53 unresolved questions extracted from the limitations and future-work sections of 322 Image Retrieval and Classification Techniques papers in our library. Each links back to the study that raised it.

What the literature leaves open

  • Large intra-class variations. Small inter-class differences in samples within the set captured in unconstrained environments. Limited performance of existing ISC methods in few-shot image set classification tasks.

    DCSCR: a class-specific collaborative representation based network for few-shot image set classification · 2026 · DOI

  • The rapid growth of video content has made efficient video retrieval a critical challenge. Traditional approaches rely heavily on manual tagging and metadata generation, which are time-consuming, inconsistent, and not scalable. The lack of efficient video retrieval systems that can handle large-scale video data.

    Semantic Video Discovery Using Deep Feature Fusion And Automated Metadata Generation · 2026 · DOI

  • There is a need for a data model to add structured data to natural history specimen images on Wikimedia Commons. The existing data model does not support a wider range of Wikimedia Commons queries.

    Wikimedia Commons data model for Natural History specimen images · 2026 · DOI

  • Limited generalization to unseen classes. Complex distributions and long-tailed, cross-domain differences. Heavy dependence on manual annotation.

    Knowledge Graph Enhanced for Zero-Shot Semantic Segmentation in Remote Sensing Imagery · 2026 · DOI

  • Visual similarity among architectural styles. Varying illumination conditions. Lack of labelled datasets.

    Deep Learning Framework for Indian Heritage Site Classification and Virtual Tour Generation using Transfer Learning · 2026 · DOI

  • The paper identifies a gap in the literature, which has traditionally focused on central place theory, neglecting the importance of gateways. The study highlights the need to consider both central place and gateway developments in understanding the rise of large centers.

    GATEWAYS: SLOW RECOGNITION BUT IRRESISTIBLE RISE · 1983 · DOI

  • The threshold θ for cosine similarity matching was set conservatively to prioritize precision over recall, but no systematic analysis of threshold selection or optimization methodology is provided.

    AI-Driven Guest Identification and Photo Retrieval System · 2026 · DOI

  • Testing was limited to four events with maximum 5,000 photos and over 100 guests; performance at significantly larger scale (thousands of guests or tens of thousands of photos) is untested.

    AI-Driven Guest Identification and Photo Retrieval System · 2026 · DOI

  • The study uses a fixed 70-30 train-test split without justification or ablation study. Cross-validation strategies, different data split ratios, and their impact on BGP-Model robustness across batik motif categories are not explored.

    Batik Motif Recognition Using the BGP-Model: A Hybrid GLCM-PCA Approach with Machine Learning Classifiers · 2026 · DOI

  • Misclassification patterns are noted (e.g., Parangsloboh misclassified as Parangklitih) but root cause analysis is absent. Systematic investigation of which GLCM texture features fail to discriminate between visually similar batik motifs would inform feature engineering improvements.

    Batik Motif Recognition Using the BGP-Model: A Hybrid GLCM-PCA Approach with Machine Learning Classifiers · 2026 · DOI

  • There are many security, performance, scalability, and architectural design issues when it comes to implementing such APIs in web apps. Most of the applications designed for academic or research purposes are very much experimental minded.

    Unsplash FindAWall: Secure, Scalable and Responsive Image-Search Platform · 2026 · DOI

  • The lack of paired data can make it challenging to ensure high-quality translations and maintain semantic coherence in the output images. Obtaining paired datasets can be impractical for many applications, as it requires manual annotation of images.

    Unpaired image-to-image translation with content preserving perspective: a review · 2026 · DOI

  • Future research could investigate the application of other quantitative metrics for image quality assessment. The development of more advanced AI image generation tools could lead to further research on the evaluation of their outputs.

    Analysis of Digital Image Quality Between Conventional Design Outputs and Generative Artificial Intelligence Images Using Digital Image Processing Methods · 2026 · DOI

  • The increasing adoption of AI image generation tools has created a need for objective methods to evaluate the quality of their outputs. There is a gap in the development of quantitative tools for image quality assessment.

    Analysis of Digital Image Quality Between Conventional Design Outputs and Generative Artificial Intelligence Images Using Digital Image Processing Methods · 2026 · DOI

  • The difficulties in obtaining power defect samples. The dominance of normal samples. The reliance on large-scale data of multi-modal large models.

    Research on the Construction of Multimodal Large Models and Self-supervised Learning for Panoramic Perception of Power Equipment · 2026 · DOI

  • The need for intelligent automated methods to distinguish relevant posts from irrelevant ones. The challenge of filtering relevant urban improvement content from the overwhelming volume of social media data.

    Multimodal detection of urban improvement indicators in public visual-text content using deep learning · 2026 · DOI

  • The requirement for extensive manual labeling and significant computing power is a major barrier to the implementation of deep learning approaches. The lack of user-friendly and accessible tools for non-AI specialists is a significant challenge.

    IAMAP: Unlocking Deep Learning in QGIS for non-coders and limited computing resources · 2026 · DOI

  • Existing zero-shot semantic segmentation frameworks are limited by their reliance on manual annotation and closed-world assumptions. The large number of classes and complex distributions make comprehensive annotation virtually impossible.

    Knowledge Graph Enhanced for Zero-Shot Semantic Segmentation in Remote Sensing Imagery · 2026 · DOI

  • There is a need for a framework that can model the formal visual language of contemporary Inner Mongolian painting. Earlier content-based retrieval systems usually described paintings through low-level visual cues, but there is a need for more advanced methods.

    Multimodal Representation Learning and Visual Analytics for Modeling the Formal Visual Language of Contemporary Inner Mongolian Painting · 2026 · DOI

  • The underutilization of modern Multimodal Large Language Models in the Digital Humanities. The limited scalability and insufficient real-time adaptability of most existing intelligent multimodal systems.

    Designing Intelligent Multimodal Assistants for Digital Humanities: A Comparative Study of Models, Modalities, and Domains · 2026 · DOI

  • Complex backgrounds often led to incomplete or misleading text extraction. Norwegian characters such as Ø, AE, and Å were not consistently recognised, even after modifying the OCR configurations. Irrelevant text extraction remained a recurring challenge, particularly when multiple products appeared within a single cropped image.

    Automated product and price information extraction from retail promotional flyers using YOLO and OCR · 2026 · DOI

  • To improve the performance of the approach on complex backgrounds and Norwegian characters. To apply the approach to other retail analytics and automated document processing applications. To evaluate the performance of the approach on a larger dataset.

    Automated product and price information extraction from retail promotional flyers using YOLO and OCR · 2026 · DOI

  • Existing traditional ISC methods classify image sets based on raw pixel features. Deep ISC methods learn deep features but fail to adaptively adjust the features when measuring set distances.

    DCSCR: a class-specific collaborative representation based network for few-shot image set classification · 2026 · DOI

  • The dataset was small and imbalanced. Some monuments had fewer than 10 samples. The system requires further development for multilingual support and AR/VR support.

    Deep Learning Framework for Indian Heritage Site Classification and Virtual Tour Generation using Transfer Learning · 2026 · DOI

  • The need for a system that integrates image restoration and caption generation. The challenge of restoring degraded images and generating accurate captions.

    DenoiseCap: A Diffusion-Based Image Restoration and Caption Generation System · 2026 · DOI

Most-cited papers in Image Retrieval and Classification Techniques

Most recent work

Find a gap in your own Image Retrieval and Classification Techniques sub-topic

This page shows what the Image Retrieval and Classification Techniques literature already flags as unresolved. To narrow it to your specific question, run the guided finder — it searches the gap library on demand and checks candidates against 250M+ OpenAlex works.

Open the Research Gap Finder →

Related topics in Computer Science

Online Learning and Analytics
851 open questions
Digital literacy in education
518 open questions
AI in Service Interactions
517 open questions
Music Technology and Sound Studies
423 open questions
Teaching and Learning Programming
384 open questions
Quantum Computing Algorithms and Architecture
368 open questions
Virtual Reality Applications and Impacts
317 open questions
Intelligent Tutoring Systems and Adaptive Learning
298 open questions
53 open questions have been extracted from the limitations and future-work passages of 322 Image Retrieval and Classification Techniques papers in our library. Each one below links back to the study that raised it, so you can read the original claim in context.

Tools for your next paper

Gaps & Literature SearchFind open questions in any field — 120,000+ open gaps mined from our own library.Pre-CheckIs your paper ready? A Tier 1–5 acceptance probability in seconds.Journal FinderA ranked shortlist of target journals from a 17,500-venue index.Duplicate Publication CheckerCheck prior-publication & salami-slicing risk across 8 sources.Citation GeneratorFreeA DOI or title → APA, MLA, Vancouver, Chicago, BibTeX, RIS. No signup.Graphical Abstract MakerFreeTurn your findings into a colorblind-safe graphical abstract.

Compare the category — Honest roundups of the AI research tools, ours listed alongside the alternatives.

Command palette

Jump anywhere, run any action.