A Systematic Review of Explainable CNNs for Image Classification

A comprehensive survey of recent XAI papers published in top-tier computer vision and AI conferences and journals (2020-2025)

130
Papers Reviewed
7
Taxonomy Dimensions
6
Years (2020-2025)
11
Datasets

Abstract

Explainable AI (XAI) is crucial for fostering human trust in deep neural network (DNN) predictions, particularly in tasks like image classification. Multiple surveys exist on XAI methodologies, however, the practical usability and reproducibility of these methods remain largely unexplored.

This paper addresses this gap by conducting a comprehensive survey of recent XAI papers published in top-tier computer vision and AI conferences and journals (supplemented with a systematic Scopus search). We categorize these works along 7 taxonomy dimensions (including a new "Type" dimension), identify prevalent datasets and evaluation metrics, and analyze the associated code repositories. A novelty of this work is its systematic analysis of the code repositories accompanying surveyed papers. The results are noteworthy: ~95% of codebases are research prototypes, and two-thirds show inconsistencies with their publications. This helps explain why benchmarking XAI methods remains so difficult in practice.

XAI Taxonomy

Scope

Local Global

Local explanations focus on specific instances, while global explanations describe overall model behavior.

Usage

Post-hoc (PH) Self-Interpretable (SI)

Post-hoc methods explain trained models; self-interpretable models are inherently transparent.

Aim

Deliberative (DE) Counterfactual (CF)

Deliberative methods justify predictions; counterfactual methods explain "why not" scenarios.

Applicability

Model-Agnostic (MA) Model-Specific (MS)

Model-agnostic methods work with any model; model-specific methods are tailored to architectures.

Methodology

Backpropagation (BP) Perturbation (PER) Prototype (PRO) Rule-based (RB) Others (OT)

Different technical approaches to generate explanations.

Modality

Visual (VS) Textual (TX)

Visual explanations show heatmaps/saliency maps; textual explanations provide descriptions.

Type

Feature-based (FBE) Concept-based (CBE) Contrastive/CF (CFE) Sample-based (SBE)

Feature-based methods highlight pixels; concept-based use high-level attributes; contrastive/counterfactual compare alternatives; sample-based use prototypes.

Reviewed Papers

Commonly Used Datasets

These 11 publicly available datasets are the most commonly used for evaluating XAI methods in reviewed papers.

XAI Methods for Comparison

Core set of methods frequently employed in state-of-the-art comparative studies.

Evaluation Metrics

Quantitative metrics used to evaluate XAI methods in computer vision.

Code Repositories Analysis

Analysis of available code repositories for XAI methods, including repository statistics and code quality assessment.

About This Survey

Key Contributions

  • ✓ Curated selection of foundational XAI works
  • ✓ Reviewed and categorized 130 recent papers from leading A* conferences and journals
  • ✓ New "Type" taxonomy dimension (Feature-based, Concept-based, Contrastive/CF, Sample-based)
  • ✓ Systematic Scopus search of top journals supplementing conference papers
  • ✓ Identified prevalent datasets, methods, and evaluation metrics
  • ✓ Code repository analysis of 81 repositories
  • ✓ Interactive website for exploring findings

Venues Covered

CVPR ECCV ICCV NeurIPS AAAI ICML SIGKDD IJCAI ICLR ICDM SIGIR CHI ACL WWW WACV

Journals (Scopus Search)

IEEE TPAMI IJCV IEEE TIP CVIU Pattern Recognition Medical Image Analysis Computational Visual Media Image and Vision Computing IEEE TCSVT IEEE TMM Signal Processing: Image Communication IEEE TMI IEEE TNNLS Neural Networks Pattern Recognition Letters IEEE RA-L IJRR JVCIR Machine Vision and Applications Expert Systems

Authors

Vahidin Hasić and Senka Krivić

Faculty of Electrical Engineering, University of Sarajevo, Bosnia and Herzegovina

Contact: vahidin.hasic@etf.unsa.ba

Paper

This interactive review accompanies the published paper:

A Systematic Review of Explainable Convolutional Neural Networks for Image Classification
Vahidin Hasić and Senka Krivić — Expert Systems, Vol. 43, No. 7, e70279, 2026.

Cite This Work

@article{hasic2026xai,
  author  = {Hasić, Vahidin and Krivić, Senka},
  title   = {A Systematic Review of Explainable Convolutional Neural Networks for Image Classification},
  journal = {Expert Systems},
  volume  = {43},
  number  = {7},
  pages   = {e70279},
  year    = {2026},
  doi     = {10.1111/exsy.70279},
  url     = {https://onlinelibrary.wiley.com/doi/abs/10.1111/exsy.70279},
  note    = {e70279 5145725}
}