Method for characterising waste discharged into a pit and associated characterisation system

Abstract

The present invention relates to a method for characterising waste discharged into a pit, comprising the following steps: capturing (110) images as a stream of waste is unloaded; detecting (180) recyclable waste in at least some of the images and, for each of these images, determining a first characteristic value characterising the presence of recyclable waste in the image and a second characteristic value characterising the waste stream; and, based on the set of first and second characteristic values, determining (190) a proportion of recyclable waste in the discharged waste stream.

Description

[0001] TITLE: Process for characterizing waste dumped in a pit and associated characterization system

[0002] The present invention relates to a method for characterizing waste dumped in a pit.

[0003] The present invention also relates to a characterization system implementing such a method.

[0004] More specifically, the invention lies in the field of characterization of waste dumped into pits of energy recovery units, called UVE.

[0005] As is well known, waste providers generally declare the nature of the waste stream to be deposited in the waste-to-energy plant's pit. This ensures the traceability of incinerated waste and allows for billing of customers based on the type of waste delivered.

[0006] However, suppliers sometimes provide a flow that is at least partially different from the one declared. This can pose a number of problems for waste-to-energy plants.

[0007] In particular, waste-to-energy furnaces are frequently blocked by oversized items, leading to unscheduled shutdowns. The risk of accidents is higher when such events occur because personnel who have to intervene do so in dangerous conditions.

[0008] Furthermore, waste streams are not billed in the same way depending on their nature, which can lead to lost revenue for waste-to-energy plants or at least a lack of traceability of inputs.

[0009] Finally, waste collectors sometimes dump significant quantities of recyclable materials, indicating a failure to comply with the source separation obligations in force since French Law No. 2015-992 of August 17, 2015, concerning the energy transition for green growth (LTECV). Therefore, waste management companies (UVEs) would like to be able to identify these non-compliant collectors and share this information with their clients, the local authorities. Local authorities are interested in this information so they can take action at the source of waste production to achieve their objectives of reducing waste and improving sorting performance. The waste management companies also wish to identify those who bring in excessively large waste so they can remove it before it is incinerated, and also to raise awareness and potentially apply penalties to these collectors.Finally, knowing the nature of the waste delivered allows not only for billing suppliers at the correct price, but also for assessing the overall lower heating value (LHV) of the incinerated waste. Errors in declarations made by suppliers skew these analyses, which are crucial for operating the facility. This is why waste-to-energy plants are interested in knowing the actual nature of the waste streams being discharged.

[0010] According to methods known in the state of the art, it is possible to characterize waste using artificial intelligence with cameras installed along the conveyor belt on which the waste travels.

[0011] State-of-the-art methods propose in particular to identify recyclable objects on the carpet by identifying these objects on the different images by making correlations between these images.

[0012] Other state-of-the-art methods involve using spectroscopy or other optical methods to characterize the nature of the waste.

[0013] It is therefore understandable that the state of the art offers complex and difficult-to-implement techniques for tracing waste.

[0014] In particular, these methods require significant computing power and / or complex optical installations along the carpet.

[0015] Furthermore, state-of-the-art methods do not allow for the resolution of the aforementioned problem, namely assessing the quantity of recyclable waste in a waste stream in order to raise awareness among offending contributors.

[0016] The present invention aims to remedy these drawbacks and to propose means of assessing the quantity of recyclable waste in a waste stream which can be implemented in a particularly simple way and without the need for significant computing power.

[0017] To this end, the invention relates to a method for characterizing waste dumped in a pit, comprising the following steps:

[0018] - filming during the unloading of a waste stream;

[0019] - detection of recyclable waste on at least some of the views and for each of these views, determination of a first characteristic value characterizing the presence of this recyclable waste on this view and a second characteristic value characterizing the waste flow;

[0020] - determination of a rate of recyclable waste in the waste stream dumped from the set of first and second characteristic values.

[0021] With these features, the invention makes it possible to determine the percentage of recyclable waste in the waste stream in a particularly simple way. Indeed, the invention proposes to process the data view by view in a decoupled manner to detect recyclable waste in each of these views. Thus, it is not necessary to track this recyclable waste on the different views, which generally requires very high computing power.

[0022] It has been shown that the rate of recyclable waste thus identified is sufficiently reliable to be communicated to the contributors.

[0023] The method according to the invention can be implemented in a particularly simple manner in UVEs because only a camera would be sufficient to ensure the necessary image capture to implement the method.

[0024] Furthermore, the images captured by the camera can be processed remotely or locally after they are taken by a computer that does not require excessive computing power. For example, a server equipped with a graphics card capable of processing several images per second can be used.

[0025] According to some embodiments, the process further includes a step of categorizing the presence of recyclable waste in the waste stream according to a predetermined class, chosen based on the rate of this recyclable waste in this waste stream.

[0026] Thanks to these characteristics, the recyclable waste rate can be categorized into one of the predetermined classes. The number of these classes can vary and can, for example, be between 2 and 10. Advantageously, the number of predetermined classes is either 2 or 4.

[0027] Categorizing the recycling rate allows for a more general classification of the amount of recyclable waste within a waste stream. As demonstrated, this rate categorization provides more valuable information to waste collectors than the exact recycling rate itself. Indeed, the latter is not always necessary to address the collectors' concerns and can present greater uncertainty compared to the rate categorization. Therefore, the reliability of the process can be particularly high, even with minimal implementation resources.

[0028] In some embodiments, the determination of the rate of recyclable waste includes the determination of a first sum corresponding to the sum of the first characteristic values ​​and a second sum corresponding to the sum of the second characteristic values.

[0029] Thanks to these characteristics, the recyclable waste rate can be defined simply as a ratio between the first sum and the second sum.

[0030] In some embodiments, the detection of recyclable waste on a corresponding view includes detecting at least one recyclable item on that view and delimiting the item(s) by a shape. Thanks to these features, the detection of recyclable objects can be implemented simply using, for example, one of the techniques already available in the field, such as supervised learning.

[0031] The shape delimiting each recyclable item can include, for example, a simple shape such as a rectangle. In some cases, this shape can be more complex and reproduce, for example, the contours of the detected recyclable waste.

[0032] In some embodiments, the first characteristic value determined on a corresponding view is the sum of the areas of the shapes delimiting recyclable waste on that view.

[0033] Thanks to these characteristics, the first characteristic value is easily calculated by summing the areas of the shapes delimiting the recyclable waste. This feature is particularly advantageous when the camera is installed opposite the waste unloading area.

[0034] In some embodiments, each second characteristic value presents an area of ​​the waste stream on the corresponding view or a generalized area of ​​the waste stream determined from the views preceding the corresponding view or from all views, the generalized area corresponding to one of the elements chosen from the group comprising:

[0035] - the sum of the areas of the waste flow on the corresponding views;

[0036] - the average of the waste flow areas on the corresponding views;

[0037] - the maximum value of the waste flow areas on the corresponding views;

[0038] - the median value of the waste flow areas on the corresponding views.

[0039] The second characteristic value can be calculated simply by calculating the areas of shapes delimiting the waste stream. Such a shape could include, for example, one or more rectangular shapes formed during the unloading of the waste stream from a truck.

[0040] In some examples, the second characteristic value representing the area of ​​the waste stream in the corresponding view is advantageous when little recyclable waste is detected in that view. This allows the recyclable waste stream to be delimited effectively, and its area can therefore be calculated accurately.

[0041] The second characteristic value presenting an averaged area or any other generalized area may be advantageous when a large number of recyclable waste items are detected on a given view.

[0042] In such a case, estimating the area of ​​the waste stream can be difficult, and estimates of recyclable waste areas can then bias the calculation. In such a case, the averaged area calculated, for example, on a predetermined number of previous views or on all views taken during unloading, provides more reliable information.

[0043] According to some embodiments, the process further includes a step of detecting the start of an unloading and / or the end of this unloading, the start of the unloading being detected following the detection of a predetermined minimum number of unloading objects and the end of the unloading being detected following the absence of detection of a minimum number of unloading objects.

[0044] Thanks to these features, it is possible to automatically detect the start or end of unloading. Therefore, no human intervention is required to initiate the estimation of the recyclable waste rate at the beginning of unloading and then stop this estimation.

[0045] In some embodiments, the process further includes a step of isolating a waste stream in the case of multiple unloadings.

[0046] Thanks to these characteristics, it is possible to distinguish several waste streams when, for example, several unloading operations are carried out in parallel.

[0047] Thus, it is possible to associate the unloading rate calculated by the process with the supplier corresponding to the waste stream.

[0048] In some embodiments, the process further includes determining the waste category for each recyclable waste item, with a rate of recyclable waste then being calculated for each category of recyclable waste.

[0049] Thanks to these characteristics, it is possible not only to determine the overall rate of recyclable waste but also the rate of this waste within each category. Thus, the final analysis provided to the waste depositor can be refined by specifying the rate for each waste category.

[0050] For the purposes of the invention, recyclable, as defined by ISO 14021, means the characteristic of products, packaging or associated components which can be taken from the waste stream by available processes and programs, and which can be collected, treated and put back into use in the form of raw materials or products.

[0051] Advantageously, each category of recyclable waste is chosen from a predetermined group including, for example, cardboard, plastic, and wood. However, the invention is not limited to this list given by way of example and may include any other recyclable material or object.

[0052] In some embodiments, the process further includes a step for detecting an unwanted object and alerting a user following this detection. These features make it possible to warn a user, such as an operator in charge of unloading control, when a bulky object is dumped into the pit. In this case, the object can be removed from the pit, for example with a grapple, to prevent subsequent blockages of the furnaces.

[0053] The invention also relates to a computer program comprising software instructions which, when executed by a computer, implement the process as defined above.

[0054] The present invention also relates to a system for characterizing waste dumped in a pit, comprising technical means configured to implement the process as defined above.

[0055] The invention will be better understood upon reading the following description, given solely by way of non-limiting example and made with reference to the accompanying drawings in which:

[0056] - [Fig. 1] Figure 1 is a schematic view of an unloading installation;

[0057] - [Fig. 2] Figure 2 is a schematic view of a characterization system according to the invention, the characterization system being connected to the unloading installation of Figure 1 by a computer network;

[0058] - [Fig. 3] Figure 3 is a flowchart of a characterization process according to the invention, the characterization process being implemented by the characterization system of Figure 2;

[0059] - [Fig. 4] [Fig. 5] Figures 4 and 5 are different views illustrating an implementation of the characterization process of Figure 3.

[0060] Figure 1 illustrates an unloading installation 10 present for example in an energy recovery unit, also called an ERU.

[0061] With reference to this figure 1, the unloading facility 10 includes a pit 12, an unloading location 13 and camera means 14.

[0062] Pit 12, for example, is a well-known type of pit and is designed to receive waste. Pit 12 can, for instance, have an elongated shape along a pit axis X perpendicular to the plane shown in Figure 1. Advantageously, pit 12 is suitable for receiving waste of all kinds, including non-recyclable waste.

[0063] Thus, for example, this pit 12 can be connected by a conduit to one or more furnace(s) of the energy recovery unit.

[0064] The unloading area 13 has one or more unloading bays suitable for unloading waste into the pit 12. For example, each unloading bay is suitable for a truck 16 or other means of transport capable of carrying the waste. This bay is then adapted so that a stream of waste 17 can fall directly into the pit 12 during unloading. In some cases, a ramp may be provided to guide the stream of waste 17 during unloading into the pit 12.

[0065] In the case of several unloading places, these places are for example arranged parallel to each other along the axis of pit X.

[0066] The image capture means 14 include one or more camera(s) positioned opposite the unloading location 13 to allow image capture of each waste stream 17 during its unloading into the pit 12.

[0067] For example, the camera system 14 has a camera positioned opposite each unloading bay. In other words, these cameras can be positioned along the axis of pit X.

[0068] By view, we mean each image taken by the camera(s) in isolation or each image from a series of images (i.e. video) taken by these cameras.

[0069] The image capture means 14 allow the images taken to be transmitted to a categorization system 20 via, for example, a computer network 19.

[0070] The computer network 19 can include any network known in itself. Thus, this network 19 can form a local network connecting the image capture means 14 to the characterization system 20 locally, or a global computer network (such as the internet for example) which allows the image capture means 14 to be connected to the characterization system 20 located remotely.

[0071] Alternatively, the shooting means 14 can be connected to the characterization system 20 by any other technically possible means such as, for example, cables.

[0072] The characterization system 20 is illustrated in more detail in Figure 2.

[0073] Thus, with reference to this figure 2, the characterization system 20 comprises an input module 21, a processing module 22, a detection module 24 and an output module 23.

[0074] Each of these modules 21 to 24 is implemented, for example, at least partially in the form of software.

[0075] In such a case, the characterization system 20 further includes memory for storing such software and a processor for executing this software. Alternatively or in addition, at least one of these modules 21 to 24 includes, at least partially, a programmable logic circuit such as an FPGA (Field-Programmable Gate Array).

[0076] The characterization system 20 thus forms, for example, a server consisting of one or more computer(s). This characterization system 20 can therefore be located, for example, locally in the corresponding energy recovery unit or remotely from it.

[0077] The input module 21 allows external data to be received and transmitted to the processing module 22 and the detection module 24.

[0078] In particular, the input module 21 is connected via the computer network 19 to the image capture means 14 to receive the views taken by these means 14. The input module 21 is also connected to a database 25 providing an object recognition model which will be explained in more detail later.

[0079] The processing module 22 allows the data received by the input module 21 to be processed in order to determine a rate of recyclable waste in a waste stream and possibly to categorize this rate into a predetermined class.

[0080] Recyclable waste refers to any waste that is worthwhile to recycle. Each recyclable waste item can, for example, be classified into a predetermined category of recyclable waste, such as cardboard, wood, and plastics. Each category may, for example, be defined by current legislation.

[0081] The detection module 24 enables the detection of objects on the views provided by the shooting means 14 using the object recognition model, as will be explained in more detail later.

[0082] The output module 23 allows the recyclable waste rate determined by the processing module 22 or the categorization of this rate according to a predetermined class to be transmitted to any interested external system.

[0083] Thus, for example, the processing module 23 can transmit this information in the form of a report 30 delivered to the waste provider.

[0084] According to some embodiments, the output module 23 is also adapted to communicate with a communication interface 32 accessible for example to an operator operating the unloading installation 10.

[0085] In particular, the output module 23 allows alerts to be transmitted to this interface 32, as will be explained in more detail later.

[0086] The characterization system 20 enables the implementation of a process for characterizing waste dumped into pit 12, which will henceforth be explained with reference to Figure 3, which presents a flowchart of its steps. During a step 110, implemented, for example, throughout the operation of the unloading installation 10, the input module 21 advantageously receives, in real time, the images taken by the imaging devices 14.

[0087] These views are for example taken continuously by the camera or each camera that is part of the shooting means 14.

[0088] In the next step 120, the input module 21 transmits these views to the detection module 24 which then implements the object recognition model.

[0089] As explained previously, the object recognition model, for example, is derived from database 25 and can detect objects of different types using a supervised or unsupervised learning algorithm. This learning algorithm has been previously trained to recognize these objects.

[0090] In particular, the learning algorithm was previously trained to recognize a waste stream and recyclable waste within that waste stream. In some cases, the learning algorithm was also previously trained to categorize each recyclable waste item into one of the predetermined recycling categories.

[0091] Of course, the use of other object recognition algorithms is also possible.

[0092] When the detection module 24 detects during this step 120 a minimum number of objects dumped into pit 12 from at least one of the unloading places within a predetermined time window, the detection module 24 concludes that there is an unloading in progress during step 130. The start of unloading is therefore detected.

[0093] In some embodiments, a spatial position filter is used to detect the start or end of unloading. This filter eliminates static detections, such as those caused by objects already present in the pit that could distort the detection.

[0094] Otherwise, when no dumped object is detected on any of the unloading places or the number of detections is not satisfactory, the detection module 24 concludes that there is no unloading in progress and then continues to process the images acquired by the input module 21 during step 110.

[0095] During step 140, the detection module 24 stores the views of the ongoing unloading process, for example, in the characterization system's RAM 20. This step 140 is implemented as long as the detection module 24 detects an ongoing unloading process during step 130. When the detection module 24 no longer detects a sufficient number of objects, for example, within a predetermined time window, it concludes that there is no longer an ongoing unloading process. The end of the unloading process is then detected, and the processing module 22 proceeds to execute the following steps 160 to 200, which form a post-processing (PP) phase for the views stored during step 140.

[0096] Alternatively, the start and end of unloading are detected by any other means (using, for example, presence sensors) or are then communicated manually, for example, by the operator.

[0097] In some embodiments, the process may further include at least one step implemented on the fly, for example during the execution of steps 130 and 140.

[0098] Thus, for example, the process may include a step 150 in which the detection module 24 analyzes each object detected in step 130 and, when it is a bulky object—that is, an object whose dimensions exceed predetermined thresholds—issues an alert to the operator. In such a case, this alert is transmitted by the output module 23 to the corresponding interface 32. Alternatively, step 150 is one of the steps implemented during the post-processing phase (PP).

[0099] In such a case, the warning given to the operator is then subsequent to the unloading.

[0100] In addition, in some embodiments, steps 160 to 200 explained below can also be implemented on the fly, for example during the execution of steps 130, 140.

[0101] In the following description, these steps 160 to 200 will be considered as belonging to the post-processing phase PP.

[0102] During an initial step 160 of the PP post-treatment phase, the processing module 22 first isolates the views relating to the same waste stream when several unloadings have been carried out in parallel.

[0103] To achieve this, various techniques are possible.

[0104] For example, when a camera is dedicated to each of the unloading places, the processing module 22 isolates the views taken by the same camera and thus considers that all these views are related to the same waste stream.

[0105] When this is not possible, or when the views captured by at least some of the cameras contain multiple waste streams within the same view, the processing module 22 delineates each group of objects within the views. This delineation can be performed, for example, using a predetermined shape such as a rectangle. When the shapes are sufficiently separated, the processing module 22 concludes that they represent different waste streams.

[0106] Otherwise, processing module 22 concludes that the two groups are part of the same flow and unites these groups by the same shape.

[0107] In another embodiment, a statistical method can be applied to determine that a set of points corresponds to a mixture of two distributions. This can be done, for example, by clustering. Each of these distributions then corresponds to a waste stream.

[0108] It is assumed that the following steps 170 to 200 are implemented for each waste stream identified by the treatment module 22.

[0109] In particular, during step 170, which may be optional in some embodiments, the processing module 22 identifies the provider of the corresponding waste stream.

[0110] This can be done, for example, by identifying on the corresponding views a registration plate of the means of transport 16 which brought back this flow of waste.

[0111] Alternatively, identification can be done using any other technically possible means.

[0112] In the next step 180, the processing module 22 processes view by view to detect recyclable waste on these views.

[0113] According to one embodiment, to do this, the processing module 22 processes each view acquired by the image capture means 14 in relation to the corresponding waste stream.

[0114] According to another embodiment, the processing module 22 processes only certain views taken by the shooting means 14. These views to be processed are, for example, taken from the set of views taken by these shooting means 14 at a predetermined frequency.

[0115] To detect recyclable waste, the processing module 22 uses the object recognition model explained previously.

[0116] In one possible embodiment, when the processing module 22 detects no recyclable waste in a given view, it rejects that view from any further analysis. Conversely, when the processing module 22 detects at least one recyclable waste item in a given view, it first determines a primary characteristic value characterizing the presence of recyclable waste in that view. In another embodiment, all views are included in the subsequent analysis. Advantageously, this primary characteristic value corresponds to the sum of the areas of all recyclable waste detected in the corresponding view.

[0117] More specifically, to achieve this, the processing module 22 delimits each recyclable waste item on the given view by a predetermined shape. This shape might, for example, be a rectangle. Then, the processing module 22 determines the area of ​​this shape and then the sum of all the determined areas. Optionally, in some examples, these areas may be normalized.

[0118] Then, the processing module 22 determines a second characteristic value characterizing the waste flow on the corresponding view.

[0119] As in the case of the first characteristic value, the second characteristic value can describe an area of ​​the corresponding waste stream.

[0120] This second characteristic value can be determined differently depending on the case.

[0121] In particular, when relatively little recyclable waste is detected on the corresponding view, the processing module 22 can then delimit the waste stream by a predetermined shape such as a rectangle and then determine the area of ​​this shape.

[0122] When, on the contrary, a large number of recyclable waste items are detected on the corresponding view so that it is not possible or difficult to determine the area of ​​the shape delimiting the corresponding waste stream, the processing module 22 determines an averaged area of ​​the waste stream.

[0123] For its part, this averaged area can also be determined according to different possibilities.

[0124] According to one possibility, this averaged area presents a moving averaged area over, for example, a predetermined number of consecutive views processed previously.

[0125] According to a second possibility, this averaged area is calculated from a predetermined number of views relating to this waste stream or from all views of this stream.

[0126] In particular, in the latter case, the second characteristic value can only be determined once for all views.

[0127] In another embodiment, a generalized area is calculated from the corresponding views. Besides the averaged area, this generalized area can correspond to:

[0128] - the sum of the areas of the waste flow on the corresponding views;

[0129] - the maximum value of the waste flow areas on the corresponding views;

[0130] - the median value of the waste stream areas on the corresponding views. In some embodiments, each first and / or each second characteristic value is / are determined with a confidence interval allowing the recyclable waste rate in the stream to be bracketed.

[0131] An example of the implementation of this step 180 is illustrated in figures 4 and 5.

[0132] Thus, this figure 4 represents five consecutive views processed by the processing module 22 of the same waste stream.

[0133] In this example, no recyclable waste was detected in views 1, 3, and 5. Therefore, these views are excluded from any future consideration.

[0134] Regarding views 2 and 4, the second characteristic value is considered to be the same for both views and is determined, for example, by averaging the area of ​​the waste stream detected across the different views. This average value is, for example, 5. Alternatively, at least in view 2, the second characteristic value V2 is determined from that same view as the area of ​​the shape delimiting the stream.

[0135] Furthermore, the first characteristic value V1 determined in relation to view 2 is equal to 3 and this value V1 determined in relation to figure 4 is equal to 1.

[0136] In the example in Figure 5, no recyclable waste is detected on view 1. On the other hand, on views 2, 3 and 4, a large number of recyclable wastes have been detected so that the first characteristic value V1 on view 2 is equal to 4, on view 3 is equal to 3 and on view 4 is equal to 1.

[0137] On the other hand, for each of these views, the second characteristic value V2 is taken as a moving average value which is equal in the example of Figure 5 to 5 in relation to each of the views 2, 3 and 4.

[0138] Many other examples are also possible.

[0139] In particular, when, for example, the first view processed by processing module 22 contains only recyclable waste (e.g., a large cardboard box), the second characteristic value can be equal to the first characteristic value. An averaged value can then be used for all consecutive views.

[0140] In the next step 190, the processing module 22 determines a rate of recyclable waste in the dumped waste stream using the first and second characteristic values ​​determined in the previous step.

[0141] Specifically, in this step, the processing module 22 calculates this rate as a ratio between the sum of the first characteristic values ​​and the sum of the second characteristic values. In the following step 200, the processing module 22 categorizes the recyclable waste rate calculated in the previous step into a predetermined class. To do this, the processing module 22 compares this rate with several thresholds defining these predetermined classes and then categorizes the resulting rate into a class. In one example, four classes are determined: class 1 non-existent, class 2 low, class 3 high, and class 4 significant. In another example, two classes are determined: class 1 non-existent and class 2 high.

[0142] Thus, for example, when the recycling rate is less than 10%, it is classified as class 1 (non-existent). When this rate is between 10% and 50%, it is classified as class 2 (low). When this rate is between 50% and 90%, it is classified as class 3 (significant). Finally, when this rate is greater than 90%, it is classified as class 4 (significant).

[0143] Of course, other classifications are still possible.

[0144] It should also be noted that, during step 180, it is also possible to associate each recyclable waste item with a category. Each category is, for example, chosen from the group including: cardboard, plastic, and wood.

[0145] Thus, in such a case, it is possible to determine the first characteristic value for each type of recyclable waste. Therefore, in such a case, during step 190, a rate for each category of recyclable waste can then be calculated.

[0146] Similarly, in step 200, each rate is categorized into the corresponding class.

Claims

DEMANDS 1. Process for characterizing waste dumped in a pit (12), comprising the following steps: - taking (110) views during the unloading of a waste stream (17); - detection (180) of recyclable waste on at least some of the views and for each of these views, determination of a first characteristic value characterizing the presence of this recyclable waste on this view and a second characteristic value characterizing the waste stream (17); - determination (190) of a rate of recyclable waste in the waste stream (17) discharged from the set of first and second characteristic values.

2. Method according to claim 1, further comprising a step (200) of categorizing the presence of recyclable waste in the waste stream (17) according to a predetermined class, chosen according to the rate of this recyclable waste in this waste stream (17).

3. A method according to claim 1 or 2, wherein the determination of the rate of recyclable waste includes the determination of a first sum corresponding to the sum of the first characteristic values ​​and a second sum corresponding to the sum of the second characteristic values.

4. A method according to any one of the preceding claims, wherein the detection of recyclable waste on a corresponding view comprises the detection on this view of at least one recyclable waste item and the delimitation of the item or items of recyclable waste by a shape.

5. A method according to claim 4, wherein the first characteristic value determined on a corresponding view is the sum of the areas of the shapes delimiting the recyclable waste on that view.

6. A method according to any one of the preceding claims, wherein each second characteristic value presents an area of ​​the waste stream (17) on the corresponding view or a generalized area of ​​the waste stream (17) determined from the views preceding the corresponding view or from all the views, the generalized area corresponding to one of the elements selected from the group comprising: - the sum of the areas of the waste flow on the corresponding views; - the average of the waste flow areas on the corresponding views; - the maximum value of the waste flow areas on the corresponding views; - the median value of the waste flow areas on the corresponding views.

7. A method according to any one of the preceding claims, further comprising a step of detecting the start of an unloading and / or the end of such unloading, the start of the unloading being detected following the detection of a predetermined minimum number of unloading objects and the end of the unloading being detected following the absence of detection of a minimum number of unloading objects.

8. A method according to any one of the preceding claims, further comprising an isolation step (160) of a waste stream (17) in the case of multiple discharges.

9. A method according to any one of the preceding claims, further comprising determining for each recyclable waste the waste category, a rate of recyclable waste then being calculated for each category of recyclable waste.

10. A method according to any one of the preceding claims, further comprising a step (150) of detecting an unwanted object and warning a user following such detection.

11. Computer program comprising software instructions which, when executed by a computer, implement the method according to any one of the preceding claims.

12. System for characterizing (20) waste dumped into a pit (12), comprising technical means (21, 22, 23, 24) configured to implement the process according to any one of claims 1 to 10.

Images