Label | EPFL Graph Search

A label (as distinct from signage) is a piece of paper, plastic film, cloth, metal, or other material affixed to a container or product, on which is written or printed information or symbols about the product or item. Information printed directly on a container or article can also be considered labelling. Labels have many uses, including promotion and providing information on a product's origin, the manufacturer (e.g., brand name), use, safety, shelf-life and disposal, some or all of which may be governed by legislation such as that for food in the UK or United States. Methods of production and attachment to packaging are many and various and may also be subject to internationally recognised standards. In many countries, hazardous products such as poisons or flammable liquids must have a warning label. Uses Labels may be used for any combination of identification, information, warning, instructions for use, environmental advice or advertising. They may be stickers, perm

Automation of the diagnosis process in the railway system : Detection of defects in concrete sleepers using vision-based machine learning models

Linah Charif

inspectors that walk over the track and check the defects on the rail surface, fasteners and sleepers. In the case of concrete sleepers, rail inspectors classify defects according to their size and occurrence over 20 sleepers. The manual inspection is error prone, it is time consuming and it can be sometimes risky. In an eort to overcome these issues, SBB has developed a project called AISI (Articial Intelligence Streckeninspection) in order to automatize the rail inspection by collecting and analysing images with diagnosis trains equipped with cameras. The objective of this thesis is integrated within the AISI Project, since its goal is the conception of a deep learning model to detect and classify concrete sleepers defects using computer vision and more specically, object detection. The defects studied in this thesis are cracks and akings (or spallings). In the literature, many researchers have tackled the subject of defect detection in concrete, especially in civil infrastructures. The most common methods a combination of image-processing and Machine Learning (image pre-processing and a classier) whereas the latest methods use deep learning (CNN, FCN and RPN, YOLO). The proposed methodology uses a Fast-RCNN model developed with Tensor ow Object Detection API. The evaluation of the model is given on a total of 996 defects and its precision and recall are given. The model shows average results as the number of false positives is high. It has been noted that the labels provided needed to be improved. The proposed approach has a great potential of development since a large amount of training data can be added thanks to the produced relabelling process. Also, the designed pipeline establishes a base that can be adjustable for various algorithms and congurations.

2021

Study of Cell Adhesion and Tracking Methods for in Vivo Cell Localisation in a Tissue-Engineering Scaffold

Thaïs Cléa Laura Séverine Clouet, Tanja Cloé Hausherr

Bone fracture healing problems, because of the loss of tissue regeneration, can affect many old or multi-pathologic patients around the world. Among the therapeutic strategies, bone substitutes, such as poly-(L-lactitve acid) with 5% wt. % ß-tricalcium phosphate (PLA/5%ß-TCP) scaffolds, are promising as a clinical solution. One strategy to tune their osteoinductive properties is the addition of bone progenitor cells. Therefore, cell- free and cell-seeded PLA/5% ß-TCP scaffolds were implanted into femoral condyles of rats to promote bone regeneration. The in vivo study showed that the bone growth inside a cell-seeded scaffold with human bone progenitor cells was delayed compared to cell-free scaffold. What happened to the cells? Did they trigger an immune response, migrate or die? Therefore, the goals of this project are to evaluate the cell adhesion in vitro to the scaffold and to propose a reliable cell labeling method to track them in future in vivo studies. In the first aim, the adhesion of the cells was qualitatively assessed by the cell morphology and the presence of fibronectin using fluorescence imaging. To quantify their adherence, seeded scaffolds were washed up to 3 times with a 0.9 % NaCl solution. The luminescence of the scaffolds and the washing solutions was measured using a cell viability assay. These experiments demonstrated that after 72h of culture, the measured adherent bone progenitor cells on the scaffolds decreased after 3 washes down to 50 % while it stayed stable after 14 days of culture. In the second aim, we focused on methods to label the cells either once implanted or prior to implantation. The first method was to translate the in situ hybridization protocol with the human specific Alu sequence from paraffin embedded samples to resin embedded ones. The obtained results were so far inconclusive. Concerning the labeling of the cells before implantation, the transmembrane label PKH26 was investigated. Its staining efficacy was high just after labeling, while it decreased drastically over time due to cell division.

2015

Synthesis of superposition images for watches, valuable articles and publicity

Sylvain Chosson, Roger Hersch

The present invention aims at synthesizing superposition images formed either by band moire shapes or by shape level lines for making the information forwarded by valuable articles or by time pieces such as watches and clocks more dynamic, as well as for improving their attractiveness and aesthetics. A further application is publicity. For synthesizing band moiré images, the present invention relies on a band moiré image layout model allowing to obtain the layout of the base the band grating, given the layouts of the band moire image and of the revealing line grating. Base and revealing layer layouts may be conceived to create band moiré image shapes whose patterns move e.g. radially, circularly, or according to a spiral trajectory. Shape level lines occur in a superposition image when e.g. a base layer comprising modified sets of lines is superposed with a revealing layer comprising a line grating. Such a base layer embeds a shape elevation profile generated from an initial motif shape image (e.g. typographic characters, words of text, symbols, logo, ornament). By moving the revealing layer in superposition with the base layer, shape level lines move dynamically between the initial motif shape boundaries and shape foreground centers, respectively shape background centers, thereby growing and shrinking. The movement of the shape level lines creates visually attractive pulsing motif shapes, e.g. a pulsing heart or pulsing text. Categories of embodiments comprise (1) visually attractive articles having moving parts (watches, clocks, vehicles, publicity display devices, fashion clothes), (2) articles such as cosmetics, drugs, perfumes and wines, where one part is moved in respect to a second part, e.g. bottles having a lid or labels composed of two layers, (3) articles where the base layer and the revealing line grating are separated by a gap and form a fixed composed layer, and (4) articles where at least one of the layers is an electronic display.