CN117218181A - Garbage volume and quality estimation method based on AI video analysis technology - Google Patents
Garbage volume and quality estimation method based on AI video analysis technology Download PDFInfo
- Publication number
- CN117218181A CN117218181A CN202311179434.8A CN202311179434A CN117218181A CN 117218181 A CN117218181 A CN 117218181A CN 202311179434 A CN202311179434 A CN 202311179434A CN 117218181 A CN117218181 A CN 117218181A
- Authority
- CN
- China
- Prior art keywords
- garbage
- volume
- weight
- module
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 57
- 238000004458 analytical method Methods 0.000 title claims abstract description 32
- 238000005516 engineering process Methods 0.000 title claims abstract description 27
- 238000004364 calculation method Methods 0.000 claims abstract description 39
- 230000008859 change Effects 0.000 claims abstract description 15
- 238000013135 deep learning Methods 0.000 claims abstract description 14
- 239000002699 waste material Substances 0.000 claims abstract description 13
- 238000003062 neural network model Methods 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims description 33
- 238000012544 monitoring process Methods 0.000 claims description 21
- 238000004140 cleaning Methods 0.000 claims description 12
- 238000012937 correction Methods 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 11
- 206010000117 Abnormal behaviour Diseases 0.000 claims description 6
- 230000010354 integration Effects 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000011897 real-time detection Methods 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 5
- 230000000694 effects Effects 0.000 claims description 4
- 238000005259 measurement Methods 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 230000005856 abnormality Effects 0.000 claims description 3
- 238000009825 accumulation Methods 0.000 claims description 3
- 230000006399 behavior Effects 0.000 claims description 3
- 238000013480 data collection Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 3
- 238000011038 discontinuous diafiltration by volume reduction Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 3
- 238000005286 illumination Methods 0.000 claims description 3
- 239000000463 material Substances 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 5
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011179 visual inspection Methods 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the technical field of garbage disposal, in particular to a garbage volume and quality estimation method based on an AI video analysis technology. The garbage volume and quality estimation method based on the AI video analysis technology obtains real-time information of a garbage site through a garbage clearance vehicle recognition model based on deep learning and OCR, a garbage type and in-out object recognition model based on deep learning, a volume estimation prediction model based on a 3D-NeRF neural network model, a normalized garbage stacking stage quality and volume calculation module, a garbage shipping stage calculation module and a vision-assisted wagon balance sensor, and estimates the weights of barreled garbage and bulk garbage respectively by combining a wagon balance average method and a volume transportation method. On the basis of considering factors such as garbage truck loading capacity, garbage density change before and after loading and the like, the method can calculate the mass and the volume of garbage more accurately, thereby facilitating the work arrangement of garbage clearing and treatment and reducing the waste of manpower and treatment resources.
Description
Technical Field
The invention relates to the technical field of garbage treatment, in particular to a garbage volume and quality estimation method based on an AI video analysis technology.
Background
At a garbage collection station of a business or a residential site, or a garbage collection point of a street at night, garbage is normally placed by a garbage can, but in many cases, because of a large number or improper disposal of garbage, there are cases where a large number of garbage bags are scattered and piled up at the garbage collection point, and in such cases, it is difficult for a worker to estimate the volume and weight of garbage to be processed, and in addition, even for garbage that has been already loaded into the garbage can, it is difficult to estimate the volume and weight thereof to a high degree because there may be a certain gap between different kinds of garbage.
When the existing garbage is estimated before transportation treatment, manual visual inspection or simple wagon balance is adopted to measure the garbage in the garbage can, the measurement mode has larger errors in volume estimation and quality estimation, and some garbage can be compressed and crushed, so that misjudgment in the volume estimation is easy to cause, the urban garbage yield is increased year by year along with the acceleration of the urban garbage process, higher requirements are put on the garbage treatment capacity, and the traditional garbage treatment mode often cannot accurately estimate the volume and weight of the garbage, so that the resource waste in the transportation and treatment process is caused; in view of this, a garbage volume and quality estimation method based on AI video analysis technology is proposed.
Disclosure of Invention
The invention aims to provide a garbage volume and quality estimation method based on an AI video analysis technology, so as to solve the problems in the background technology.
In order to achieve the above purpose, the present invention provides the following technical solutions: a garbage volume and quality estimation method based on AI video analysis technology comprises the following steps:
s1, identifying a garbage collection vehicle by using a garbage collection vehicle identification model based on deep learning and OCR technology;
s2, identifying the garbage types and the in-out objects by utilizing a garbage type and in-out object identification model based on deep learning;
s3, estimating the garbage volume by adopting a volume estimation prediction model based on a 3D-NeRF neural network model;
s4, calculating the mass and the volume of the garbage stacking stage by normalizing the mass and the volume calculating module of the garbage stacking stage;
s5, calculating data of the garbage shipment stage by using a garbage shipment stage calculation module;
s6, measuring the weight of the garbage by using a vision-assisted wagon balance sensor.
Preferably, the garbage collection vehicle recognition model based on deep learning and OCR comprises the following steps:
s101, data collection: collecting pictures of the dry garbage truck, the wet garbage truck and the box garbage truck from multiple angles and under illumination conditions, and collecting pictures of corresponding license plates and other vehicle identifications so as to provide basic data for training an OCR model;
s102, data preprocessing and marking: carrying out preprocessing operations such as denoising, graying, size adjustment and the like on the collected pictures, and then marking targets such as garbage trucks, license plates and the like in the pictures so as to be ready for training Yolo 7 and OCR models;
s103, training a Yolo 7 model: and training a Yolo 7 model by using the preprocessed and marked pictures, so that the dried garbage truck, the wet garbage truck and the box garbage truck can be identified. In the training process, the model performance is optimized by methods of adjusting model parameters, increasing training times and the like;
s104, training an OCR model: training an OCR model by using the license plate and other vehicle identification pictures acquired in the S101, and optimizing the model performance by adjusting model parameters, increasing training times and other methods in the training process;
s105, model integration and optimization: integrating the Yolo 7 model and the OCR model after training in the S103 and the S104, when the Yolo 7 identifies the garbage truck, the OCR model identifies license plates and other identifications, and carries out weighted fusion on output results of the two models;
s106, actual application and optimization: the integrated model is deployed into an actual scene, such as a garbage disposal plant, a garbage disposal station and the like, and then further optimized according to the actual effect.
Preferably, the garbage type and in-out object recognition model based on deep learning comprises the following modules:
garbage type recognition module: the module can accurately identify different types of garbage such as dry garbage, wet garbage, bulk garbage, construction garbage and recyclable garbage through information comparison;
personnel monitoring module: the module can monitor the personnel entering and exiting conditions in the garbage disposal area in real time and help manage the garbage disposal area;
the area change detection module: the module can monitor the state of the garbage treatment area by analyzing the area change before and after the personnel enter and exit, and can realize the real-time tracking of the activities such as garbage carrying, classifying, stacking and the like;
abnormal behavior alarm module: when the module recognizes abnormal behaviors such as garbage disposal without classification according to regulations, unauthorized access to forbidden areas and the like, an alarm is automatically sent out, and the safety of the garbage disposal areas is improved;
auxiliary garbage cleaning module: the module can provide data support for garbage cleaning work by monitoring the garbage stacking condition in real time, and improves garbage cleaning efficiency.
Preferably, the volume estimation prediction model based on the 3D-NeRF neural network model comprises the following modules:
a garbage volume estimation module: the module realizes accurate estimation of the garbage volume through measurement and calibration of the ground area and analysis of the visible coverage area;
and the real-time monitoring module is used for: the module processes the images captured by the camera in real time, timely discovers backlog and clearance requirements, and improves garbage disposal efficiency;
garbage type recognition module: the module can identify different kinds of garbage by combining with an image identification technology, and provides support for volume estimation of different garbage;
and an auxiliary decision module: the module can provide data support for garbage treatment and clearing decisions through estimating the garbage volume, so that the garbage treatment efficiency and accuracy are improved;
an abnormality alarm module: the module combines a real-time monitoring function, and gives an alarm when abnormal accumulation or illegal behaviors are identified, so that the safety of a garbage disposal area is improved.
Preferably, the mass and volume calculation module for the normalized garbage stacking stage comprises the following calculation steps:
s201, monitoring a garbage stacking area in real time: real-time monitoring object changes in the garbage stacking area by using a garbage type based on Yolo 7 and an entering object recognition model;
s202, identifying the garbage type and the garbage volume: measuring and calibrating the ground shooting angle and the ground shooting height of the camera by using a 3D NeRF space model, so as to calculate the garbage stacking volume in real time;
s203, integrating wagon balance sensor data: integrating the identified garbage type, volume and other information with data from the wagon balance sensor to provide more comprehensive garbage information;
s204, calculating the weight of the newly added garbage: calculating the weight of newly added garbage according to the integrated data, and providing accurate weight data;
s205, recalculating and calibrating the whole garbage volume and weight: and (3) recalculating and calibrating the volume and the weight of the whole garbage to obtain real-time garbage volume and weight data so as to make more accurate garbage treatment decisions.
Preferably, the calculating step code of the mass and volume calculating module in the normalized garbage stacking stage is as follows:
def monitor_garbage_area():
while True:
real-time detection of refuse dump object using Yolo 7 model
detected_objects = detect_objects()
Calculation of garbage volume using 3D NeRF model #
garbage_volume = calculate_volume(detected_objects)
# acquisition of wagon balance sensor data
scale_data = get_scale_data()
# integration of detected objects, calculated volume and wagon balance data
integrated_data = integrate_data(detected_objects, garbage_volume, scale_data)
# calculating the weight of the newly added garbage according to the volume and the type
new_garbage_weight=calculate_weight(integrated_data)
# update overall garbage volume and weight
update_total_volume_weight(new_garbage_weight)。
Preferably, the garbage shipment stage calculation module comprises the following calculation steps:
s301, identifying a garbage collection vehicle: determining the type and the in-out time of the garbage truck by using a garbage truck identification model based on Yolo 7 and OCR;
s302, identifying and counting the garbage to be cleared: presetting specific identification and counting according to different garbage clearing events, such as the number of wet garbage cans, dry garbage cans, bulk garbage bags and tipping buckets of building garbage tipping vehicles;
s303, summarizing the amount of the garbage to be cleared: after the garbage collection vehicle leaves, summarizing the garbage quantity;
s304, calculating the volume and the weight of the residual garbage: calculating the volume and the weight of the reserved garbage according to the collected garbage collection quantity;
s305, error correction: error correction is performed according to empirical data to improve the accuracy of the estimation.
Preferably, the calculation code of the garbage shipment stage calculation module is as follows:
def monitor_garbage_removal():
while True:
identifying garbage collection vehicles using Yolo 7 and OCR models
garbage_truck_info = detect_garbage_truck()
if garbage_truck_info is not None:
# identifying and counting refuse for cleaning
counted_garbage=count_garbage(garbage_truck_info)
# aggregate garbage collection and transportation quantity
summarized_garbage=summarize_garbage(counted_garbage)
Calculation of the volume and weight of the refuse remaining
remaining_volume_weight=calculate_remaining_volume_weight(summarized_garbage)
Error correction based on empirical data
corrected_data=error_correction(remaining_volume_weight)。
Preferably, when the garbage type recognition module recognizes that the garbage type is a dry garbage can and a wet garbage can, the mass estimation calculation code is as follows:
# initialize total_weight to 0
total_weight = 0
# initialize total_count to 0
total_count = 0
# for each garbage_station in garbage_stations:
for garbage_station in garbage_stations:
# add weight scale data of the garbage station
add_weight_scale(garbage_station)
# for each garbage_record in garbage_station.data:
for garbage_record in garbage_station.data:
# if the record only contains garbage:
if is_garbage_only(garbage_record):
# add the weight to the total weight
total_weight += garbage_record.weight
# increment the total count
total_count += 1
# calculate the average weight of barreled garbage
average_weight_of_barreled_garbage = total_weight / total_count。
preferably, when the garbage type recognition module recognizes that the garbage type is bulk garbage, the mass estimation adopts the following calculation method:
s401, real-time detection: real-time monitoring is carried out on the garbage site by using an AI video analysis technology, and the whole volume of bulk garbage is obtained;
s402, estimating volume change: by observing the volume reduction of the garbage truck after one shipment, and the time of garbage workers carrying garbage, the number of shipments of the garbage truck per day, plus the overall capacity of the garbage truck, a formula is formed:
bulk waste weight = Σ (volume change x number of shipments x waste cart capacity)/(waste worker handling time x 3;
s403, calculating the total weight: assuming that the average three shipments of the garbage truck are equivalent to the overall one-time shipment, the volume and density of each three shipments are accumulated, and the total weight of the bulk garbage is calculated.
Compared with the prior art, the invention provides a garbage volume and quality estimation method based on an AI video analysis technology, which has the following beneficial effects:
1. the garbage volume and quality estimation method based on the AI video analysis technology combines the AI video analysis technology, a wagon balance averaging method and a volume and traffic method, and realizes accurate estimation of the weights of barreled garbage and bulk garbage. The garbage site information is collected in real time, and the change of the loading capacity of the garbage truck and the density of the garbage before and after the loading is considered, so that the mass and the volume of the garbage can be calculated more accurately. Thereby optimizing the work arrangement of garbage disposal and reducing the consumption of manpower and processing resources.
2. According to the garbage volume and quality estimation method based on the AI video analysis technology, the average loading capacity of the garbage truck and the change of the garbage density before and after loading are analyzed, so that the garbage quality and volume estimation method can adapt to various scene changes, the applicability of the estimation method is improved, and further, the AI video analysis technology is utilized to monitor the garbage site in real time so as to ensure the accuracy of calculated data when the garbage enters and exits.
3. According to the garbage volume and quality estimation method based on the AI video analysis technology, the AI video analysis technology is utilized to collect data information of garbage trucks, dry garbage, wet garbage and the like, and the garbage treatment efficiency is continuously improved through calculation and comparison.
Drawings
FIG. 1 is a schematic diagram of a deployment state of the present invention;
FIG. 2 is a comparison chart of the south-north dry garbage data of the invention;
FIG. 3 is a diagram showing a comparison of statistics and project reporting data of a street system according to the present invention;
FIG. 4 is a comparison chart of fashion garbage data in a business district according to the present invention;
FIG. 5 is a diagram showing the comparison of statistics and project report data of a garbage system during a certain business district according to the present invention;
FIG. 6 is a comparison chart of wet garbage data of a certain street according to the present invention;
FIG. 7 is a diagram showing statistics of a wet garbage system and project report data of a street according to the present invention;
FIG. 8 is a graph comparing fashion wet garbage data from a business district according to the present invention;
FIG. 9 is a diagram showing statistics of a wet garbage system and project report data for a certain business district according to the present invention;
FIG. 10 is a schematic view of the construction waste status of the Tian-Di-nan area of the siphon bridge according to the present invention;
FIG. 11 is a graph of statistical data of the construction waste system in the Tian-Di-nan area of the siphon bridge according to the present invention.
Detailed Description
As shown in fig. 1, the present invention provides a technical solution: a garbage volume and quality estimation method based on AI video analysis technology comprises the following steps:
s1, identifying a garbage collection vehicle by using a garbage collection vehicle identification model based on deep learning and OCR technology;
s2, identifying the garbage types and the in-out objects by utilizing a garbage type and in-out object identification model based on deep learning;
s3, estimating the garbage volume by adopting a volume estimation prediction model based on a 3D-NeRF neural network model;
s4, calculating the mass and the volume of the garbage stacking stage by normalizing the mass and the volume calculating module of the garbage stacking stage;
s5, calculating data of the garbage shipment stage by using a garbage shipment stage calculation module;
s6, measuring the weight of the garbage by using a vision-assisted wagon balance sensor.
In an embodiment of the invention, a deep learning and OCR based garbage collection vehicle recognition model comprises the steps of:
s101, data collection: collecting vehicle pictures of different types (such as a dry garbage truck, a wet garbage truck and a box garbage truck) including pictures of different angles and under different illumination conditions, and collecting pictures of corresponding license plates and other vehicle identifications at the same time so as to provide basic materials for training an OCR model;
s102, data preprocessing and marking: preprocessing the pictures acquired in the step S101, including denoising, graying, size adjustment and the like, and then marking targets such as garbage trucks, license plates and the like in the pictures to prepare for training Yolo 7 and OCR models;
s103, training a Yolo 7 model: training a Yolo 7 model by using the preprocessed and marked pictures in the S102, so that different types of garbage trucks can be identified, and in the training process, the model performance can be optimized by methods of adjusting model parameters, increasing training times and the like;
s104, training an OCR model: training an OCR model by using the license plate and other vehicle identification pictures acquired in the S101, and optimizing the model performance by adjusting model parameters, increasing training times and other methods in the training process;
s105, model integration and optimization: integrating the Yolo 7 model and the OCR model after training in the S103 and the S104, when the Yolo 7 identifies the garbage truck, the OCR model identifies license plates and other identifications, and carries out weighted fusion on output results of the two models;
s106, actual application and optimization: the integrated model is deployed into an actual scene, such as a garbage disposal plant, a garbage disposal station and the like, and then further optimized according to the actual effect.
Specifically, the garbage type and in-out object recognition model based on deep learning comprises the following modules:
garbage type recognition module: the module can identify different kinds of garbage through information comparison, including dry garbage cans, wet garbage cans, bulk garbage, construction garbage and recyclable garbage;
personnel monitoring module: the module can monitor the personnel entering and exiting conditions in the garbage disposal area in real time and help manage the garbage disposal area;
the area change detection module: the module can monitor the state of the garbage disposal area by analyzing the area change before and after the personnel enter and exit;
abnormal behavior alarm module: when the module recognizes abnormal behaviors such as garbage disposal without classification according to regulations, unauthorized access to forbidden areas and the like, an alarm is automatically sent out, and the safety of the garbage disposal areas is improved;
auxiliary garbage cleaning module: the module can provide data support for garbage cleaning work by monitoring the garbage stacking condition in real time, and improves garbage cleaning efficiency.
In addition, the volume estimation prediction model based on the 3D-NeRF neural network model comprises the following modules:
a garbage volume estimation module: the module realizes accurate estimation of the garbage volume through measurement and calibration of the ground area and analysis of the visible coverage area;
and the real-time monitoring module is used for: the module processes the images captured by the camera in real time, timely discovers backlog and clearance requirements, and improves garbage disposal efficiency;
garbage type recognition module: the module can identify different kinds of garbage by combining with an image identification technology, and provides support for volume estimation of different garbage;
and an auxiliary decision module: the module can provide data support for garbage treatment and clearing decisions through estimating the garbage volume, so that the garbage treatment efficiency and accuracy are improved;
an abnormality alarm module: the module combines a real-time monitoring function, and gives an alarm when abnormal accumulation or illegal behaviors are identified, so that the safety of a garbage disposal area is improved.
Specifically, the mass and volume calculation module for the normalized garbage stacking stage comprises the following calculation steps:
s201, monitoring a garbage stacking area in real time: real-time monitoring object changes in the garbage stacking area by using a garbage type based on Yolo 7 and an entering object recognition model;
s202, identifying the garbage type and the garbage volume: measuring and calibrating the ground shooting angle and the ground shooting height of the camera by using a 3D NeRF space model, so as to calculate the garbage stacking volume in real time;
s203, integrating wagon balance sensor data: integrating the identified garbage type, volume and other information with data from the wagon balance sensor to provide more comprehensive garbage information;
s204, calculating the weight of the newly added garbage: calculating the weight of newly added garbage according to the integrated data, and providing accurate weight data;
s205, recalculating and calibrating the whole garbage volume and weight: and (3) recalculating and calibrating the volume and the weight of the whole garbage to obtain real-time garbage volume and weight data so as to make more accurate garbage treatment decisions.
Meanwhile, the calculation step codes of the mass and volume calculation module in the normalized garbage stacking stage are as follows:
def monitor_garbage_area():
while True:
real-time detection of refuse dump object using Yolo 7 model
detected_objects = detect_objects()
Calculation of garbage volume using 3D NeRF model #
garbage_volume = calculate_volume(detected_objects)
# acquisition of wagon balance sensor data
scale_data = get_scale_data()
# integration of detected objects, calculated volume and wagon balance data
integrated_data = integrate_data(detected_objects, garbage_volume, scale_data)
# calculating the weight of the newly added garbage according to the volume and the type
new_garbage_weight=calculate_weight(integrated_data)
# update overall garbage volume and weight
update_total_volume_weight(new_garbage_weight)。
In an embodiment of the invention, the steps of the garbage shipment stage calculation module are as follows:
s301, identifying a garbage collection vehicle: determining the type and the in-out time of the garbage truck by using a garbage truck identification model based on Yolo 7 and OCR;
s302, identifying and counting the garbage to be cleared: presetting specific identification and counting according to different garbage clearing events, such as the number of wet garbage cans, dry garbage cans, bulk garbage bags and tipping buckets of building garbage tipping vehicles;
s303, summarizing the amount of the garbage to be cleared: after the garbage collection vehicle leaves, summarizing the garbage quantity;
s304, calculating the volume and the weight of the residual garbage: calculating the volume and the weight of the reserved garbage according to the collected garbage collection quantity;
s305, error correction: error correction is carried out according to the empirical data so as to improve the accuracy of estimation, and the calculation codes of the garbage shipment stage calculation module are as follows:
def monitor_garbage_removal():
while True:
identifying garbage collection vehicles using Yolo 7 and OCR models
garbage_truck_info = detect_garbage_truck()
if garbage_truck_info is not None:
# identifying and counting refuse for cleaning
counted_garbage=count_garbage(garbage_truck_info)
# aggregate garbage collection and transportation quantity
summarized_garbage=summarize_garbage(counted_garbage)
Calculation of the volume and weight of the refuse remaining
remaining_volume_weight=calculate_remaining_volume_weight(summarized_garbage)
Error correction based on empirical data
corrected_data=error_correction(remaining_volume_weight)。
When the garbage type recognition module recognizes that the garbage types are the dry garbage bin and the wet garbage bin, the mass estimation calculation code is as follows:
# initialize total_weight to 0
total_weight = 0
# initialize total_count to 0
total_count = 0
# for each garbage_station in garbage_stations:
for garbage_station in garbage_stations:
# add weight scale data of the garbage station
add_weight_scale(garbage_station)
# for each garbage_record in garbage_station.data:
for garbage_record in garbage_station.data:
# if the record only contains garbage:
if is_garbage_only(garbage_record):
# add the weight to the total weight
total_weight += garbage_record.weight
# increment the total count
total_count += 1
# calculate the average weight of barreled garbage
average_weight_of_barreled_garbage = total_weight / total_count。
further, when the garbage type recognition module recognizes that the garbage type is bulk garbage, the mass estimation adopts the following calculation mode:
s401, real-time detection: real-time monitoring is carried out on the garbage site by using an AI video analysis technology, and the whole volume of bulk garbage is obtained;
s402, estimating volume change: by observing the volume reduction of the garbage truck after one shipment, and the time of garbage workers carrying garbage, the number of shipments of the garbage truck per day, plus the overall capacity of the garbage truck, a formula is formed:
bulk waste weight = Σ (volume change x number of shipments x waste cart capacity)/(waste worker handling time x 3;
s403, calculating the total weight: assuming that the average three shipments of the garbage truck are equivalent to the overall one-time shipment, the volume and density of each three shipments are accumulated, and the total weight of the bulk garbage is calculated.
The foregoing invention has been generally described in great detail, but it will be apparent to those skilled in the art that modifications and improvements can be made thereto. Accordingly, it is intended to cover modifications or improvements within the spirit of the inventive concepts.
Claims (10)
1. A garbage volume and quality estimation method based on an AI video analysis technology is characterized by comprising the following steps: the method comprises the following steps:
s1, identifying a garbage collection vehicle by using a garbage collection vehicle identification model based on deep learning and OCR technology;
s2, identifying the garbage types and the in-out objects by utilizing a garbage type and in-out object identification model based on deep learning;
s3, estimating the garbage volume by adopting a volume estimation prediction model based on a 3D-NeRF neural network model;
s4, calculating the mass and the volume of the garbage stacking stage by normalizing the mass and the volume calculating module of the garbage stacking stage;
s5, calculating data of the garbage shipment stage by using a garbage shipment stage calculation module;
s6, measuring the weight of the garbage by using a vision-assisted wagon balance sensor.
2. The method for estimating the volume and the quality of garbage based on the AI video analysis technique as set forth in claim 1, wherein: the garbage collection vehicle recognition model based on deep learning and OCR comprises the following steps:
s101, data collection: collecting vehicle pictures of different types (such as a dry garbage truck, a wet garbage truck and a box garbage truck) including pictures of different angles and under different illumination conditions, and collecting pictures of corresponding license plates and other vehicle identifications at the same time so as to provide basic materials for training an OCR model;
s102, data preprocessing and marking: preprocessing the pictures acquired in the step S101, including denoising, graying, size adjustment and the like, and then marking targets such as garbage trucks, license plates and the like in the pictures to prepare for training Yolo 7 and OCR models;
s103, training a Yolo 7 model: training a Yolo 7 model by using the preprocessed and marked pictures in the S102, so that different types of garbage trucks can be identified, and in the training process, the model performance can be optimized by methods of adjusting model parameters, increasing training times and the like;
s104, training an OCR model: training an OCR model by using the license plate and other vehicle identification pictures acquired in the S101, and optimizing the model performance by adjusting model parameters, increasing training times and other methods in the training process;
s105, model integration and optimization: integrating the Yolo 7 model and the OCR model after training in the S103 and the S104, when the Yolo 7 identifies the garbage truck, the OCR model identifies license plates and other identifications, and carries out weighted fusion on output results of the two models;
s106, actual application and optimization: the integrated model is deployed into an actual scene, such as a garbage disposal plant, a garbage disposal station and the like, and then further optimized according to the actual effect.
3. The method for estimating the volume and the quality of garbage based on the AI video analysis technique as set forth in claim 1, wherein: the garbage type and in-out object recognition model based on deep learning comprises the following modules:
garbage type recognition module: the module can identify different kinds of garbage through information comparison, including dry garbage cans, wet garbage cans, bulk garbage, construction garbage and recyclable garbage;
personnel monitoring module: the module can monitor the personnel entering and exiting conditions in the garbage disposal area in real time and help manage the garbage disposal area;
the area change detection module: the module can monitor the state of the garbage disposal area by analyzing the area change before and after the personnel enter and exit;
abnormal behavior alarm module: when the module recognizes abnormal behaviors such as garbage disposal without classification according to regulations, unauthorized access to forbidden areas and the like, an alarm is automatically sent out, and the safety of the garbage disposal areas is improved;
auxiliary garbage cleaning module: the module can provide data support for garbage cleaning work by monitoring the garbage stacking condition in real time, and improves garbage cleaning efficiency.
4. The method for estimating the volume and the quality of garbage based on the AI video analysis technique as set forth in claim 1, wherein: the volume estimation prediction model based on the 3D-NeRF neural network model comprises the following modules:
a garbage volume estimation module: the module realizes accurate estimation of the garbage volume through measurement and calibration of the ground area and analysis of the visible coverage area;
and the real-time monitoring module is used for: the module processes the images captured by the camera in real time, timely discovers backlog and clearance requirements, and improves garbage disposal efficiency;
garbage type recognition module: the module can identify different kinds of garbage by combining with an image identification technology, and provides support for volume estimation of different garbage;
and an auxiliary decision module: the module can provide data support for garbage treatment and clearing decisions through estimating the garbage volume, so that the garbage treatment efficiency and accuracy are improved;
an abnormality alarm module: the module combines a real-time monitoring function, and gives an alarm when abnormal accumulation or illegal behaviors are identified, so that the safety of a garbage disposal area is improved.
5. The method for estimating the volume and the quality of garbage based on the AI video analysis technique as set forth in claim 1, wherein: the mass and volume calculation module for the normalized garbage stacking stage comprises the following calculation steps:
s201, monitoring a garbage stacking area in real time: real-time monitoring object changes in the garbage stacking area by using a garbage type based on Yolo 7 and an entering object recognition model;
s202, identifying the garbage type and the garbage volume: measuring and calibrating the ground shooting angle and the ground shooting height of the camera by using a 3D NeRF space model, so as to calculate the garbage stacking volume in real time;
s203, integrating wagon balance sensor data: integrating the identified garbage type, volume and other information with data from the wagon balance sensor to provide more comprehensive garbage information;
s204, calculating the weight of the newly added garbage: calculating the weight of newly added garbage according to the integrated data, and providing accurate weight data;
s205, recalculating and calibrating the whole garbage volume and weight: and (3) recalculating and calibrating the volume and the weight of the whole garbage to obtain real-time garbage volume and weight data so as to make more accurate garbage treatment decisions.
6. The method for estimating the volume and the quality of garbage based on the AI video analysis technique as set forth in claim 5, wherein: the calculation step codes of the mass and volume calculation module in the normalized garbage stacking stage are as follows:
def monitor_garbage_area():
while True:
real-time detection of refuse dump object using Yolo 7 model
detected_objects = detect_objects()
Calculation of garbage volume using 3D NeRF model #
garbage_volume = calculate_volume(detected_objects)
# acquisition of wagon balance sensor data
scale_data = get_scale_data()
# integration of detected objects, calculated volume and wagon balance data
integrated_data = integrate_data(detected_objects, garbage_volume, scale_data)
# calculating the weight of the newly added garbage according to the volume and the type
new_garbage_weight=calculate_weight(integrated_data)
# update overall garbage volume and weight
update_total_volume_weight(new_garbage_weight)。
7. The method for estimating the volume and the quality of garbage based on the AI video analysis technique as set forth in claim 1, wherein: the garbage shipment stage calculation module comprises the following calculation steps:
s301, identifying a garbage collection vehicle: determining the type and the in-out time of the garbage truck by using a garbage truck identification model based on Yolo 7 and OCR;
s302, identifying and counting the garbage to be cleared: presetting specific identification and counting according to different garbage clearing events, such as the number of wet garbage cans, dry garbage cans, bulk garbage bags and tipping buckets of building garbage tipping vehicles;
s303, summarizing the amount of the garbage to be cleared: after the garbage collection vehicle leaves, summarizing the garbage quantity;
s304, calculating the volume and the weight of the residual garbage: calculating the volume and the weight of the reserved garbage according to the collected garbage collection quantity;
s305, error correction: error correction is performed according to empirical data to improve the accuracy of the estimation.
8. The method for estimating the volume and the quality of garbage based on the AI video analysis technique as set forth in claim 7, wherein: the calculation codes of the garbage shipment stage calculation module are as follows:
def monitor_garbage_removal():
while True:
identifying garbage collection vehicles using Yolo 7 and OCR models
garbage_truck_info = detect_garbage_truck()
if garbage_truck_info is not None:
# identifying and counting refuse for cleaning
counted_garbage=count_garbage(garbage_truck_info)
# aggregate garbage collection and transportation quantity
summarized_garbage=summarize_garbage(counted_garbage)
Calculation of the volume and weight of the refuse remaining
remaining_volume_weight=calculate_remaining_volume_weight(summarized_garbage)
Error correction based on empirical data
corrected_data=error_correction(remaining_volume_weight)。
9. The method for estimating the volume and the quality of garbage based on the AI video analysis technique as set forth in claim 3, wherein: when the garbage type recognition module recognizes that the garbage types are dry garbage cans and wet garbage cans, the mass estimation calculation codes are as follows:
# initialize total_weight to 0
total_weight = 0
# initialize total_count to 0
total_count = 0
# for each garbage_station in garbage_stations:
for garbage_station in garbage_stations:
# add weight scale data of the garbage station
add_weight_scale(garbage_station)
# for each garbage_record in garbage_station.data:
for garbage_record in garbage_station.data:
# if the record only contains garbage:
if is_garbage_only(garbage_record):
# add the weight to the total weight
total_weight += garbage_record.weight
# increment the total count
total_count += 1
# calculate the average weight of barreled garbage
average_weight_of_barreled_garbage = total_weight / total_count。
10. the method for estimating the volume and the quality of garbage based on the AI video analysis technique as set forth in claim 3, wherein: when the garbage type recognition module recognizes that the garbage type is bulk garbage, the mass estimation adopts the following calculation mode:
s401, real-time detection: real-time monitoring is carried out on the garbage site by using an AI video analysis technology, and the whole volume of bulk garbage is obtained;
s402, estimating volume change: by observing the volume reduction of the garbage truck after one shipment, and the time of garbage workers carrying garbage, the number of shipments of the garbage truck per day, plus the overall capacity of the garbage truck, a formula is formed:
bulk waste weight = Σ (volume change x number of shipments x waste cart capacity)/(waste worker handling time x 3;
s403, calculating the total weight: assuming that the average three shipments of the garbage truck are equivalent to the overall one-time shipment, the volume and density of each three shipments are accumulated, and the total weight of the bulk garbage is calculated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311179434.8A CN117218181A (en) | 2023-09-13 | 2023-09-13 | Garbage volume and quality estimation method based on AI video analysis technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311179434.8A CN117218181A (en) | 2023-09-13 | 2023-09-13 | Garbage volume and quality estimation method based on AI video analysis technology |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117218181A true CN117218181A (en) | 2023-12-12 |
Family
ID=89036491
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311179434.8A Pending CN117218181A (en) | 2023-09-13 | 2023-09-13 | Garbage volume and quality estimation method based on AI video analysis technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117218181A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117470360A (en) * | 2023-12-27 | 2024-01-30 | 江苏迈鼎科技(集团)有限公司 | Intelligent unattended wagon balance system based on guiding |
-
2023
- 2023-09-13 CN CN202311179434.8A patent/CN117218181A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117470360A (en) * | 2023-12-27 | 2024-01-30 | 江苏迈鼎科技(集团)有限公司 | Intelligent unattended wagon balance system based on guiding |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111553902B (en) | Highway road surface safety monitoring system based on big data | |
| US8020767B2 (en) | Systems and methods for monitoring a quantity of waste in a waste transfer station environment | |
| US20190210798A1 (en) | Solid Waste Evaluation and Weighing System | |
| CN106530733A (en) | Remote bridge overload law enforcement system based on video linkage capture | |
| CN117218181A (en) | Garbage volume and quality estimation method based on AI video analysis technology | |
| CN102941864A (en) | Train loading state high-definition monitoring and overloading detection method | |
| CN110782214A (en) | Intelligent supervision system applied to construction waste recovery treatment | |
| CN113361868B (en) | Intelligent environmental sanitation receiving and transporting analysis system based on all-condition vehicle-mounted weighing technology | |
| CN112528208B (en) | Weighing-free AI intelligent recognition truck overload estimation method, device and system | |
| CN116342895B (en) | Method and system for improving sorting efficiency of renewable resources based on AI (advanced technology attachment) processing | |
| CN115527364B (en) | Traffic accident tracing method and system based on radar data fusion | |
| CN111830917A (en) | Digital anti-cheating quality inspection control system and method for raw materials of papermaking enterprise | |
| CN110633639B (en) | Receiving and transporting supervision method and receiving and transporting supervision system | |
| CN112687099A (en) | Method and device for judging overload suspected vehicle | |
| CN117689400B (en) | Overrun source tracing method and system based on data tag | |
| CN113487877A (en) | Road vehicle illegal parking monitoring method | |
| CN114758127A (en) | Urban scene garbage detection system based on big data | |
| JP7424945B2 (en) | Failure detection device, toll collection system, failure detection method, and program | |
| CN114111988B (en) | Intelligent weighing and shunting control method and system for garbage recovery | |
| CN117854016A (en) | Method and device for detecting abnormal vehicle based on expressway | |
| CN115273474B (en) | RPA patrols and examines robot and patrols and examines management system based on artificial intelligence | |
| CN116307052A (en) | Smart city garbage recycling detection system based on big data | |
| CN213482700U (en) | Digital anti-cheating quality inspection management and control system for raw materials of papermaking enterprises | |
| CN115271481A (en) | Industrial solid waste intelligent classification processing monitoring management system based on machine vision | |
| CN108460542B (en) | Grain depot warehouse-in and warehouse-out management method and device and computer readable storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |