Context: The passive monitoring of biodiversity is a cost-efficient approach for the study, management, and conservation of animal populations and communities. Here, “passive” refers to methods using autonomous sensors that minimally disturb the studied species or their habitat. Examples of passive monitoring methods include acoustic monitoring, visual monitoring (e.g., camera traps, satellite imagery), chemical analyses, and environmental DNA (eDNA). Two of the most common forms of passive monitoring are visual and acoustic monitoring. In these methods, one or more sensors are deployed at single locations or over large geographical areas. These sensors are used to detect and identify organisms at various taxonomic levels, including family, genus, and species.  Passive monitoring of nature reduce the costs and human labor time in the field. However, processing the collected data remains time consuming and the analysis would benefit from automatic methods. 

Passive Visual Monitoring (PVM) has emerged as an effective alternative for studying animal populations, enabling the detection of species presence and the monitoring of temporal changes in population dynamics.  

Scope of the book: The aim of this book is to group recent research works that use modern signal processing techniques, machine learning, and deep learning methods such as GNNs, GANs, Transformers and graph neural networks, spectral graph and hypergraph neural networks to address significant challenges related to PVM, in both terrestrial and aquatic ecosystems. The goal is to encourage engineers to build stronger collaborations with biologists and ecologists by providing them suitable and robust methods for PVM to address the challenges of understanding the terrestrial and aquatic environments. 

More specifically, the proposed book will include chapters related to recent modern relevant theoretical and practical challenges met in PVM that can be classified as follows: 

• Detecting species and individuals with few, one, or no training examples. This includes supervised few-, one-, and zero-shot learning, unsupervised learning, and reinforcement learning. 
• Insufficient labeled data due to the rarity of the observed species or impossibility to label all the data because of the large number of biological varieties. 
• Multiple-overlaped detections. Sound cacophonies or visual collage that hampers individual identification. Common in gregarious species and in situations such as food frenzy. Spatial information (direction of arrival), signal tracking (as in particle filters), and source separation (BSS) can help in these cases. 
• Multimodal techniques. These are techniques that integrate information from different types of sensors (e.g., sound, images) in order to create new information or more robust applications. 

• Acoustic event detection. Techniques for the rapid and large-scale detection of acoustic events including biological, geological, and human-made sounds.  
• Novel ways to embed or represent visual data. This includes graph deep learning embeddings for example.    

   
  The contributions address key challenges in the passive visual monitoring of biodiversity through the application of signal processing and artificial intelligence techniques. 

  How to submit: Please send a prospective title, abstract and list of authors to tbouwman@univ-lr.fr before October 30, 2025. An accompanying website will be provided to the readers in order to improve the access to the works of the contributors increasing their visibility in the community. 

  Publisher: CRC Press

  Editors: 
  Thierry BOUWMANS (IEEE Senior Member, ACM Member, Top 2% Standford), Laboratoire MIA, La Rochelle Université, France. (Thierry BOUWMANS)

• Anastasia ZAKHAROVA, Laboratoire MIA, La Rochelle Université, France. 
• Badri Subudhi, IIT Jammu, India. (Badri N Subudhi Web Page)
• Meghna Kapoor, IIT Jammu, India. (Dr. Meghna Kapoor)