AI recognition technology has entered market application. With the advancement of algorithms, AI image recognition has transformed from medical applications to factory logistics. Many factories use image recognition systems to manage warehouses, optimize traffic flows, and reduce errors. Optimistic about the huge business opportunity in the checking of goods for logistics, the startup team, WorMIT INC, in the Startup Terrace Linkou has developed an AI image recognition system. Through AI technology, a successful recognition rate of nearly 99% can be achieved, greatly reducing the errors in delivering goods.
With the rise of e-commerce and large distribution channels, more and more third-party logistics and logistics centers have been established. However, in the demand for small-quantity and diversified orders or fast logistics services, problems such as items shipped do not match with the order, wrong goods are delivered, or missing of goods are often encountered. These seemingly small problems often lead to large costs in acceptance check. Therefore, under the principle of the "Trusted Acceptance Check" (a concept first proposed by the General Manager of Retail Support International, Zhao-Sai Hong), an additional inspection operation from pallets to logistics trucks started to be valued by logistics center.
Da-Yin Zhan, Founder of WorMIT INC, said that compared with traditional barcode scanning or RFID inspection, image recognition by AI can greatly shorten the inspection time for products. If inspection is done manually or by scanning the barcodes of products, repeated scanning and acceptance check operations are required. On average, it will take approximately 5-10 minutes for each batch of product to complete its shipment inspection. However, if AI image recognition is adopted, it may only take 10 to 15 seconds to complete the inspection. How does WorMIT INC do that?
Teach AI to Recognize Images, Achieving an Accuracy of 99%
"It's very simple. We teach AI to perform image recognition." Da-Yin Zhan gave an example. A well-known room temperature logistics center needs organize nearly 800 boxes of goods every day. If there are about 800 orders in each store, an average of 10~70 boxes of different goods will be picked and placed on the same pallet, resulting in the combinations of 10~70 times 800. If the product is manually checked, it is difficult to be completed in a short period of time, not to mentioned that human eyes cannot carry out inspection operation continuously, and the chance of error is extremely high. On the contrary, if AI knows how to recognize these goods, the results can be totally different.
Da-Yin Zhan said that WorMIT INC uses AI algorithms and analysis to teach AI how to compare the shape, color, and font of each box, improving the accuracy of recognition. Each box has five sides and four different angles. By adopting the same-product overlapping recognition and dynamic single-lens recognition, even the items that are placed slantingly in a certain angle and be recognized by the AI technology developed by WorMIT. Therefore, when the pallet passes the camera, all items on the pallet can be check for compliance. Verified by the field test conducted by WorMIT INC, the recognition accuracy can reach as high as 99.9%.
Chiung-Hui Huang, Head of R&D of WorMIT INC, who has a background in logistics management said that the "Trusted Acceptance Check" is a very important niche indicator for large logistics companies. When the delivery accuracy falls within a certain percentage, the checking operation between the delivery truck driver and store personnel is no longer required. Typically, the time required for the checking operation in each store is about 5-10 minutes. By March 2020, there are 5,751 7-11 stores in Taiwan. This suggests that by adopting the new technology, nearly 500 hours can be saved per day. Considering the wage for the delivery truck driver is NT$ 200 per hour (not yet considering the wage of other personnel), a total of NT$ 30 million can be saved in a year. So far, only AI can achieve this.
Bonded Relationship through Entrepreneurship, Joining Hands to Enter the Innovative Logistics Market
To be able to develop the first AI recognition system in Taiwan, the team must have impressive background. WorMIT is a professional team composed of logistics management consultants, logistics system architects, senior engineers and product managers. They have been served in big well-known companies. In the beginning of the entrepreneurship, there were nearly 20 people in the team. However, since there were no orders for a long period of time, people left the team one after another. Eventually, only four people who still have faith stayed in the team, applying AI to innovate the logistics market in Taiwan.
Currently, WorMIT has received an order from a well-known memory company in Taiwan. Due to the end-of-the-year inventory, the company required 20 people to carry out the inventory operation. These people needed to work overtime on both weekdays and weekends. It took two months for them to complete the inventory operation. After the WorMIT system is introduced, the labor required is reduced to 8 people. Moreover, the total working period is reduced from two months to one and a half days, achieving great performance. There are currently more than 3,000 large and small logistics warehouses and material warehouses in Taiwan. Da-Yin Zhan is confident that AI systems can help logistics companies to carry out transformation and upgrade.
WorMIT INC stationed in the Startup Terrace Linkou last year and was awarded a subsidy project. It was also selected to participate in the Microsoft for Startups BizSpark Program and won the second place in the Huawei IoT Hackathon competition. The current product line also includes CAPS (Electronic Assisted Picking System) and a simultaneous recording system for shipment inspection. It has customers from China, Taiwan and Malaysia. Da-Yin Zhan emphasized that the international exchange and startup resources of the Startup Terrace Linkou are essential for the WorMIT team to bring AI recognition technology to the world.
Figure：WorMIT INC warehousing items and examples of the application of the AI recognition system
Written/Photographed by: Startup Terrace Editorial team