# Are there any 'object detectors' in the hidden layers of CNNs trained to identify objects or scenes?

arXiv: https://arxiv.org/abs/2007.01062 · HTML: https://mlsanity.com/paper/2007.01062/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 2 Jul 2020.

## Built on by (most cited first, 6 of 6)

- [Don't trust your eyes: on the (un)reliability of feature visualizations](https://mlsanity.com/paper/2306.04719/) (2023)
- [Dear XAI Community, We Need to Talk! Fundamental Misconceptions in Current XAI Research](https://mlsanity.com/paper/2306.04292/) (2023)
- [Scientific Inference With Interpretable Machine Learning: Analyzing Models to Learn About Real-World Phenomena](https://mlsanity.com/paper/2206.05487/) (2022)
- [Are Deep Neural Networks Adequate Behavioural Models of Human Visual Perception?](https://mlsanity.com/paper/2305.17023/) (2023)
- [A Survey on Deep Domain Adaptation and Tiny Object Detection Challenges, Techniques and Datasets](https://mlsanity.com/paper/2107.07927/) (2021)
- [Cluster Flow: how a hierarchical clustering layer make allows deep-NNs more resilient to hacking, more human-like and easily implements relational reasoning](https://mlsanity.com/paper/2304.14081/) (2023)

## Abstract (arXiv)

Various methods of measuring unit selectivity have been developed with the aim of better understanding how neural networks work. But the different measures provide divergent estimates of selectivity, and this has led to different conclusions regarding the conditions in which selective object representations are learned and the functional relevance of these representations. In an attempt to better characterize object selectivity, we undertake a comparison of various selectivity measures on a large set of units in AlexNet, including localist selectivity, precision, class-conditional mean activity selectivity (CCMAS), network dissection,the human interpretation of activation maximization (AM) images, and standard signal-detection measures. We find that the different measures provide different estimates of object selectivity, with precision and CCMAS measures providing misleadingly high estimates. Indeed, the most selective units had a poor hit-rate or a high false-alarm rate (or both) in object classification, making them poor object detectors. We fail to find any units that are even remotely as selective as the 'grandmother cell' units reported in recurrent neural networks. In order to generalize these results, we compared selectivity measures on units in VGG-16 and GoogLeNet trained on the ImageNet or Places-365 datasets that have been described as 'object detectors'. Again, we find poor hit-rates and high false-alarm rates for object classification. We conclude that signal-detection measures provide a better assessment of single-unit selectivity compared to common alternative approaches, and that deep convolutional networks of image classification do not learn object detectors in their hidden layers.