IWN Report — Tuesday, 29 September 2026

The Wildlife Conservation Trust (WCT) has released a free, offline species-recognition model that identifies 40 kinds of Central Indian wildlife in camera-trap photographs with 97.3 per cent accuracy, aimed at students, researchers and forest officers who have thousands of images to sort and little money to sort them with.

According to WCT, the tool was built with Addax Data Science and trained on 1.1 million camera-trap images taken at more than 10,000 separate locations, mostly in the dry and moist deciduous forests of Central India. WCT says it gathered those images through systematic camera-trap surveys carried out over several years across the landscape.

The trust’s lead for its Conservation Research – Central India Programme says models trained on global image collections already exist, but that adding region-specific photographs improves their accuracy, because local training data captures how animals in a particular landscape actually look and what sits behind them in the frame.

The software runs on a user’s own computer without an internet connection and needs no coding, and WCT describes it as open source. The trust says it can be downloaded from Addax Data Science’s AddaxAI page.

A WCT research associate said the model saves thousands of hours of manual labour, and the trust’s president said it should let students, researchers and forest officers analyse their data faster and with more confidence. WCT adds that automated identification gives consistent results with less observer bias, and lets researchers widen their study areas and time spans without hiring more people to match.

Sources: Wildlife Conservation Trust · Addax Data Science (AddaxAI)