Birdfy unveils updated AI bird ID features at June 26 webinar

Birdfy used a June 26 webinar to show off new Birdfy AI capabilities, including sex identification, animal recognition and location-based filtering. The company also said its updated model is now live after community corrections improved accuracy on hard-to-distinguish species.

Why it matters: - Birdfy’s latest AI update aims to make smart birdwatching more accurate, more educational and more useful in real backyards. - The changes also show how user-submitted corrections are helping refine species identification over time. - The company is positioning AI as a broader nature-monitoring tool, not just a bird ID system.

What happened: - Birdfy held a tech webinar on June 26 to introduce the new Birdfy AI and its latest identification capabilities. - The session was themed “Meet the Mind Behind Birdfy AI: How Bird Identification Works.” - More than 100 birders attended. - Birdfy CTO Udall Hu and Birdfy AI Consultant Alec Roseto were the speakers. - Brand Ambassador Gary Herritz hosted the event.

The details: - Hu said Birdfy OrniSense is powered by a vision-language model, or VLM, rather than a traditional AI model. - Hu described the VLM as understanding birds, habits and environment, instead of only matching pixels. - The model identifies bird species and also explains why a bird was identified that way. - The system uses smart fallback when a photo or video is too blurry to support a species-level label. - In those cases, Birdfy OrniSense returns a broader category instead of forcing a wrong guess. - The AI also evaluates habitat context, such as wetlands, deserts or icy branches, to eliminate unlikely matches. - Birdfy first launched bird AI identification for smart feeders in 2021. - Birdfy added nesting process identification for smart birdhouses in 2022. - Since Birdfy OrniSense launched, the company has rolled out three additional AI features: sex identification, animal recognition and geographic location filtering. - Birdfy products can now identify the sex of select bird species. - Roseto said the feature can help users understand which birds are visiting feeders and how behavior may change with the seasons. - The model can explain plumage differences between male and female birds in species including Brown-headed Cowbirds, House Finches and Northern Cardinals. - Birdfy OrniSense can recognize up to 16 animal species. - The supported animals include squirrels, deer, raccoons, cats and dogs. - Animal recognition can help users keep non-target wildlife away from birdseed. - The feature can also send instant alerts when a familiar squirrel appears. - Geographic location filtering narrows the species list using location data. - Birdfy said its GeoBird Database helps the AI distinguish a Black-billed Magpie in the United States from a Eurasian Magpie in Eurasia. - The location layer also improves local identification of lookalike species such as Carolina Chickadees and Black-capped Chickadees. - Roseto said accurate location data separates Carolina from Black-capped Chickadees in most cases. - Herritz said users should submit correction feedback when an ID is wrong. - Birdfy said real-world correction data is valuable for training the model. - Hu said the model previously struggled to tell Common Grackles from Brewer’s Blackbirds apart. - Hundreds of birder corrections helped the updated model distinguish between those two species. - Hu said the updated model is now officially live.

Between the lines: - Birdfy is leaning on a hybrid approach: AI model changes plus crowd-sourced corrections from users. - The webinar framed accuracy as a community project, not just a software upgrade. - The company is also expanding beyond species recognition into health, personalization and conservation use cases. - That suggests Birdfy wants the platform to become a data network for birders and citizen science.

What’s next: - Birdfy said it is training a sick bird identification feature to recognize avian diseases such as avian pox and severe feather degradation. - The app is expected to notify users when illness is detected so they can clean smart feeders. - The company is also developing a self-learning personalized AI with account-level memory. - Birdfy is researching bird tag and band recognition for automatic sightings reporting to conservation groups and scientific institutions. - Hu said leg-band IDs are difficult because the numbers are microscopic and often blurred in motion. - Hu said the team is still pushing OrniSense VLM to make that feature work. - Birdfy shared social links for Instagram, Facebook, YouTube, TikTok and X.

The bottom line: - Birdfy’s updated AI is moving from basic bird recognition toward context-aware identification, broader wildlife detection and community-trained accuracy.