How PellencST cut their time-to-model by 90% and gained a competitive edge in the sorting machine market

Context

PellencST, a pioneering manufacturer of waste sorting machines, has dramatically enhanced its market position by being one of the first to integrate Computer Vision (CV) into its machines. By leveraging Picsellia's comprehensive CV platform, PellencST not only reduced its time-to-market but also expedited the customization of CV models for specific client needs, gaining significant operational efficiency and boosting client satisfaction.

‍

• Time-to-Model Reduction: PellencST reduced the development and adaptation time for CV models by 90%
• Efficiency Gains: PellencST quadrupled the speed and efficiency of building and deploying CV models
• Competitive Advantage: The ability to develop and deploy customized solutions quicker enabled PellencST to reinforce their position as market leaders.

‍

About PellencST:

PellencST leads the way in the conception of advanced waste sorting solutions, working in close collaboration with companies ranging from environmental organizations and municipalities, to design offices, operators, and recycling leaders like Paprec, Veolia, and Suez.

The company specializes in enhancing sorting processes through innovative technologies, including AI-driven quality control to ensure precise material handling, benefiting both operational costs and environmental impact. With 2,500 machines in operation in more than 40 countries, Pellenc ST is the undisputed market leader in France, and a key player internationally.

‍

PellencST's CV Journey:

Facing a growing demand for complex sorting capabilities that couldn’t be tackled with regular sensors, like silicone cartridge sorting and improved paper/cardboard separation, PellencST embraced Computer Vision technology to address these applications. After initially externalizing the development of CV models, they made the decision to internalize and accelerate AI capabilities, which was crucial for maintaining real-time processing on their machines. The collaboration with Picsellia streamlined these efforts, enabling rapid model deployment and lifecycle management that aligned with business goals. Picsellia’s platform allows their R&D and business teams to work together on industrializing AI models. 

"Using Picsellia has revolutionized how we develop, monitor, and deploy our models, making complex CV applications more reliable and quicker to market."

‍

Build vs. Buy Decision:

PellencST evaluated the need for a swift, effective solution against the slower, resource-intensive option of in-house development. Developing an in-house solution would consume significant time and resources, requiring extensive involvement from the R&D team and possibly misaligning with business and industrialization needs. In contrast, choosing Picsellia allowed them to bypass these hurdles. 

Picsellia's business-oriented and intuitive platform integrates well with open-source tools like Airflow and MLFlow, fitting the R&D team's requirements while also meeting business needs. The platform's ability to unify all MLOps tools in one place coupled with consulting services, was crucial. This strategic decision enabled PellencST to focus on rapid deployment and secured them a competitive advantage in the waste sorting industry. 

‍

Business Impact of Computer Vision with Picsellia:

With Picsellia, PellencST quadrupled the speed and efficiency of building and deploying CV models. The platform facilitated faster response to customer requirements and contributed to securing significant contracts. The integration of CV has propelled PellencST to the forefront of the waste sorting industry, enhancing their technological capabilities and solidifying their position as a market leader.

Reduced Time-to-Market:

PellencST reduced the development and adaptation time for CV models from an initial 6-9 months down to just 1 month, including training, testing, and validation phases. This acceleration allowed them to rapidly meet client demands and capitalize on new opportunities, effectively reducing time-to-market.

Efficiency in Annotation Process:

Leveraging Picsellia’s pre-annotation capabilities, PellencST nearly halved the time required for data preparation, going from a 1-month manual annotation process to just 2 weeks. Additionally, the data acquisition process was streamlined from 1-2 months to only 3-4 days, significantly reducing the model development cycle.

Cost Reduction

By enabling efficient model fine-tuning and iteration specifically tailored to client needs, Picsellia's platform likely contributed to a decrease in labor and operational costs associated with model development and deployment.

Operational Efficiency:

The centralization of all objects (models, datasets, experiments) in a single place provided full autonomy in model operability, particularly for retraining. This centralization, combined with daily use and real-time monitoring capabilities, translates into substantial operational efficiencies.

Impact on Competitive Positioning:

The ability to quickly develop and deploy customized solutions not only reinforced PellencST's competitive edge but also directly contributed to securing major contracts. This strategic advantage, enabled by Picsellia, is crucial in a competitive market where speed and adaptability are key.

‍

Moving Forward:

With several innovative projects in the pipeline, PellencST is set to expand their use of CV technologies to enhance the sorting capabilities of their machines, so they can perform more complex tasks, like distinguishing between food and non-food wastes. Picsellia continues to support these initiatives by providing an integrated, collaborative environment that bridges the gap between R&D and production, ensuring that PellencST remains at the cutting edge of waste management technology.

‍