The old port of Cannes is a mythical marina on the French south east coast. It is a gateway to Le Suquet, a typical Provencal district with a rich architectural heritage in the center of the city and the start of the famous Boulevard de la Croisette. The port has more than ten docks and a ferry terminal. Regular connections to Saint Tropez and the islands of Lérins are provided there. More than 350,000 coastal shuttle passengers and over 300,000 cruise ship passengers pass through it each year. In addition, many pleasure boats, sailboats and luxury yachts are moored at the quays.
Nautic Spot, an integrator specializing in securing marinas, in collaboration with ACIC, proposed to the port of Cannes to carry out a Proof of Concept in order to demonstrate the effectiveness of image analysis solutions in a port environment. The port decided to give us access to a number of cameras for a month and allowed us to deploy image analysis solutions if we could bring our own server. The goal was to use as many image analysis solutions as possible in order to demonstrate its effectiveness and utility.
Among its products, ACIC has a solution to count people in a zone. This solution, called CrowdNumbering, has been proposed to SNCF. In concrete terms, this product makes it possible to determine a virtual zone and to instantly count the number of people present in this zone. The counting frequency can be customized (every minute, every 10 seconds, …). With the rig
After selecting the cameras that we found the most interesting, we got in touch with our provider Capsule Technology so that they could deploy the server. We then contacted Milestone in order to have an operating environment and the possibility of raising alarms.
In total, we equipped 14 cameras with 7 separate image analysis solution software during the months of September and October 2020. We have deployed:
- a people counting license based on an IA module
- a vehicle counting license based on an IA module
- two licenses for counting people in an area
- four line crossing detection licenses
- a loitering detection license
- two counting and mask detection licenses based on an AI module
- three camera anomaly detection licenses
The analytics events linked to the different cameras were fed back directly to Milestone during the PoC.
ht settings, the error rate is relatively low and can be below 10%. The video analytics which allows the counting is based on artificial intelligence models (deep neural networks). They have been trained to detect the appearance of a person or a head, regardless of their pose or orientation. This product runs on an analytics server and requires a GPU.
The use of ACIC image analysis solutions on the port has been successful. The data collection allowed the administration to know precisely the passage, the influx and the movements of crowds in strategic points of the port.