'Part'ly First Among Equals: Semantic Part-Based Benchmarking for State-of-the-Art Object Recognition Systems

Ravi Kiran Sarvadevabhatla, Shanthakumar Venkatraman, R. Venkatesh Babu

The research was archived in the Munich Personal RePEc Archive (MPRA).

Abstract

This paper examines Spain's rapid evolution from a cash-heavy economy to a modern "cashless society" between 1989 and 2014. By introducing a novel "cashless indicator" and applying time-series econometrics, we demonstrate how banking modernization, the introduction of the Euro, and macroeconomic growth drove the explosive adoption of electronic payments. Our results reveal a long-run decoupling of electronic transactions from physical cash holdings, tied instead to banking deposits and overall economic health.

Introduction

Historically, Spain maintained a strong cultural and economic reliance on physical cash, adopting modern payment tools like checks and ATMs significantly later than many other European nations. However, starting in the late 1980s, the country underwent a rapid modernization of its financial infrastructure. This study aims to quantify this transition and map the long-term relationships between retail payment systems, monetary aggregates (M1, M2, M3), and broader economic activity.

Methodology

To analyze the transition, we employ several econometric techniques:

  • The Cashless Indicator: A novel metric defined as the ratio of the total value of cashless transactions (CLT) to Currency Held by the Public (CHP), multiplied by 100. A value exceeding 100 signifies a "cashless society."
  • Structural Break Tests: Andrews-Ploberger and Andrews-Quandt tests are used to identify significant shifts in historical data trends without prior assumptions about the dates.
  • Cointegration Analysis: The Johansen procedure identifies stable, long-run relationships between non-stationary variables such as GDP and money supply.
  • Error Correction Models (ECMs): These models capture short-run dynamics and the speed at which the system returns to equilibrium after temporary deviations.

Data Sources

The study utilizes annual data from 1952 to 2014 for historical context, with a focused econometric analysis on the 1989–2014 period. Data on monetary aggregates (CHP, M1, M2, M3) and cashless transactions (checks, card payments, credit transfers, direct debits) were sourced from the Bank of Spain and the European Central Bank (ECB) MasterSeries BlueBook. Macroeconomic data, including GDP and inflation, were obtained from the Spanish National Statistics Institute (INE).

Key Results and Discussion

Explosive Growth and the Cashless Threshold

The value of cashless transactions in Spain grew by 788% from 1989 to 2014, averaging 9.13% annual growth. Our proposed "cashless indicator" shows that Spain officially crossed the threshold into a cashless society around 2002, with the indicator consistently staying well above 100 thereafter.

The Euro as a Catalyst

The introduction of the physical Euro in 2002 caused a massive structural break in cash holdings. This event acted as a catalyst, accelerating the shift toward electronic payments and permanently altering the relationship between the public and physical currency.

Long-Run Cointegration and Decoupling

Johansen tests confirmed statistically significant long-run relationships between cashless transactions and M1, M2, M3, and GDP. Notably, the growth of the cashless economy shows no significant relationship with physical cash (CHP). This suggests that the expansion of electronic payments is fundamentally tied to banking sector growth and national income rather than fluctuations in physical currency.

Short-Run Dynamics

Error Correction Models revealed that short-run imbalances between cashless transactions and the real economy (GDP) or liquid money (M1) trigger reliable adjustment processes. If these variables drift apart, they pull back toward their equilibrium path year over year, ensuring the stability of the payment ecosystem.

Conclusion

Spain's transition to a cashless society was driven by early investments in banking computerization and the exogenous shock of the Euro's introduction. The decoupling of electronic transactions from physical cash holdings highlights a fundamental shift in economic behavior. While the study concludes in 2014, the established trends provided the foundation for the subsequent acceleration of mobile and contactless payments seen in the late 2010s.

Summary updated

Abstract

An examination of object recognition challenge leaderboards (ILSVRC, PASCAL-VOC) reveals that the top-performing classifiers typically exhibit small differences amongst themselves in terms of error rate/mAP. To better differentiate the top performers, additional criteria are required. Moreover, the (test) images, on which the performance scores are based, predominantly contain fully visible objects. Therefore, ‘harder’ test images, mimicking the challenging conditions (e.g. occlusion) in which humans routinely recognize objects, need to be utilized for benchmarking. To address the concerns mentioned above, we make two contributions. First, we systematically vary the level of local object-part content, global detail and spatial context in images from PASCAL VOC 2010 to create a new benchmarking dataset dubbed PPSS-12. Second, we propose an object-part based benchmarking procedure which quantifies classifiers’ robustness to a range of visibility and contextual settings. The benchmarking procedure relies on a semantic similarity measure that naturally addresses potential semantic granularity differences between the category labels in training and test datasets, thus eliminating manual mapping. We use our procedure on the PPSS-12 dataset to benchmark top-performing classifiers trained on the ILSVRC-2012 dataset. Our results show that the proposed benchmarking procedure enables additional differentiation among state-of-the-art object classifiers in terms of their ability to handle missing content and insufficient object detail. Given this capability for additional differentiation, our approach can potentially supplement existing benchmarking procedures used in object recognition challenge leaderboards.
BibTeX
@inproceedings{sarvadevabhatla2016part,
  author = {Ravi Kiran Sarvadevabhatla and Shanthakumar Venkatraman and R. Venkatesh Babu},
  title = {{'Part'ly First Among Equals: Semantic Part-Based Benchmarking for State-of-the-Art Object Recognition Systems}},
  booktitle = {ACCV},
  year = {2016},
  url = {https://openreview.net/forum?id=1zAvwUelQbi},
}
Figure