Automatic detection of pig tail biting

TAIL is a diagnostic tool that automatically detects when a pig bites the tail of a fellow pig.

#1 Detect tail-biting events in pig production

The cutting of tails at the birth of piglets is prohibited as a routine practice but is still tolerated ( under veterinary advice) by European regulations.

This practice is tending to disappear, driven by a society that is increasingly concerned about animal welfare.

👉 You are a veterinarian, a technician, a breeder, and you are looking for a simple and quick tool to help you accompany changes in practice.

Good news! TAIL has been developed especially for you! 🤩

#2 A tool that detects biting events for you

TAIL detects caudophagy instead of the veterinarian or technician, in a non-intrusive way.

👉 TAIL combines a removable video camera, placed in a collective pen of post-weaning and fattening pigs, with an Artificial Intelligence and Computer Vision Cloud application that autonomously detects tail biting behaviour.

Click on the video to enlarge it

#3 No additional work required

When the analysis of animal behaviour is done manually, it means hours of video for technicians and veterinarians to watch.

This approach to observing animals is not widely used, as it involves many constraints:

Significant additional workload: Veterinarians have to watch hours of video looking for significant events in an already busy schedule.
Decrease in regularity of performance: The attention of the person viewing the videos decreases after a few minutes of viewing, making the analysis less effective on complex videos.
Difficulties in detecting significant events: Pigs are kept in collective pens (about 20-30 pigs). In high activity periods, it is very difficult to distinguish bites.

#4 Save a lot of time!

Once the videos are transmitted to our cloud analysis platform, the events searched for are automatically detected by the Artificial Intelligence algorithms and returned to the user in the form of statistics.

The process only takes 5 minutes for a vet!

A veterinarian would take an average of 3 hours to analyse 1 hour of video. On the Dilepix Cloud, the analysis takes less than 1 hour and the process takes only 5 minutes for the veterinarian (file deposit time) because he/she can continue his/her activities while the platform analyses the video for him/her.

#5 Animal welfare and the evolution of practices

TAIL helps the pig industry to support changes in practices and move towards greater animal welfare

Non-invasive: This tool does not require the installation of any sensors on the animal.
Respect for the biological cycle: The camera monitors at a distance and does not disturb the rhythm of the animals. It has no impact on their behaviour.
Animal comfort & welfare: Caudophagy results from several factors (boredom, unsuitable temperature, disturbing activity, feeding problems, etc.). Understanding its origins will eventually make it possible to establish prediction models to anticipate this phenomenon.

#6 TAIL's performance according to our expert

To develop this tool, we relied on nearly 700 videos from farms with tail biting problems. To design and train the detection neural networks, Dilepix analysed the equivalent of 230 hours of video.

To develop TAIL, we created a new type of neural network that analyses behaviour over time, taking into account the context before and after an event.

This technique significantly improves performance and avoids false positives.

Mathieu SIMON - Production Manager

Mathieu-Simon-dilepix

Difference between frame-by-frame analysis vs. time-based video analysis

Left screen: image classifications (frame by frame analysis) - Right screen: video classifications (temporal analysis)

#7 What are the benefits for the user?

Analysis results delivered within minutes

Only three steps are required:

Setup of the camera in the breeding pen
One or two days later: recovery of the camera
Recovery of the video files recorded on the SD card

Once the videos are transmitted to the Dilepix web platform, the analysis results are delivered to the user within minutes.

A simple tool for everyone

TAIL is a software and analysis platform that combines artificial intelligence and computer vision analysis.
The tool works on all types of video cameras provided that a minimum of 15 frames/second is observed.

Respecting animal welfare

Tail does not require a sensor to be fitted to the pigs
Its use does not disturb the rhythm of the animals
It allows concrete measurement of biting phenomena in order to quickly implement corrective actions.

#8 Take part in the project

Our objective is that TAIL becomes THE TOOL that helps breeders, veterinary practices, laboratories and cooperatives to support the cessation of tail docking.

To deploy this tool on a large scale and make it compatible with a maximum number of farms, we are looking for an industrial partner with expertise in pig production or breeding.

This collaboration will make it possible to act on 4 drivers:

Anticipate regulatory changes in Europe and contribute to animal welfare
Helping veterinarians and technicians to better target their advice and prescriptions
Save time in observing animals
Provide higher value advice to farmers

Talk about my project

Detecting tail biting in pigs

Anticipate the phenomena of pig caudophagy

Establish predictive models
Non-invasive and animal welfare friendly method
Save time in observing animals

Ask for more infos

Client Testimonial

A true partner relationship

"We worked with Dilepix in the SYNAPSE project. The great strength of the team is it honesty in always giving the limits of the capabilities for each development step and never overselling the results. Constant efforts are made to ensure that the objectives are realistic, achievable and constructive. We were really in a co-construction process and we are very satisfied with this collaboration".

Frédéric DIMUR - Digital Open Innovation Director - Boehringer Ingelheim

Read his interview

#8 Frequently asked questions (FAQ)

What type of camera do I need?

You need a simple monitoring camera with an SD card or a recorder (NVR type) installed above a box with a 45° angle.

How are the results reported?

The results are returned to you on our cloud analysis platform in the form of raw data (time-stamped bite sequences), which can be downloaded in CSV format or interrogated directly via our API.

How many cameras do I need?

You need one camera per box

More information

Dilepix' blog

Access to blog

35000 Rennes, FRANCE

Expertise

Products

Dilepix

Legal Notice