A deep learning-based detection method for oxygen cylinder and acetylene cylinder transportation in the same vehicle

Abstract

The present application belongs to the technical field of oil and gas well target detection and target comparison, and particularly relates to a kind of oxygen cylinder and acetylene cylinder same vehicle transportation detection method based on deep learning.The oxygen cylinder and acetylene cylinder same vehicle transportation detection method based on deep learning can be used to realize real-time detection of the irregular accident of oxygen cylinder and acetylene cylinder same vehicle transportation, not only can meet the management requirements of oilfield enterprises, but also can timely find the above-mentioned safety hazards and issue early warning to avoid accidents.The present application constructs YOLOv5 network model structure, trains neural network model, compares the coordinate positions between the detection target results of acetylene cylinder and the detection target results of transport vehicle, and the coordinate positions between the detection target results of oxygen cylinder and the detection target results of transport vehicle, and judges whether acetylene cylinder and oxygen cylinder are on the same transport vehicle based on intersection over union, and calculates the coordinate distance between the detection target results of acetylene cylinder and the detection target results of oxygen cylinder.

Description

Technical Field

[0001] This invention belongs to the field of oil and gas production safety risk target detection technology, and in particular relates to a deep learning-based detection method for transporting oxygen cylinders and acetylene cylinders in the same vehicle. Background Technology

[0002] In oil and gas well production operations, many construction scenarios require the use of acetylene and oxygen cylinders. However, based on safety production standards, acetylene and oxygen cylinders are not allowed to be transported together. This is because transporting acetylene and oxygen cylinders together can cause friction and collisions during transport, leading to significant safety hazards and even extreme accidents such as explosions and deflagrations. Therefore, for construction sites with extremely high safety requirements, such as oil companies, it is necessary to investigate the possibility of transporting oxygen and acetylene cylinders together to effectively protect the lives and property of workers and prevent accidents.

[0003] However, the inventors discovered that there is currently no detection method capable of identifying the simultaneous transport of oxygen and acetylene cylinders in the same vehicle. It is worth noting that deep learning technology, due to its potential in object detection, can be considered by engineers as one means to overcome the aforementioned technological shortcomings. Among them, the patent document titled "A Method for Detecting the Start-up and Stoppage of a Video Pumping Unit" (application number: CN202011531759.4) describes a method for detecting the start-up and stoppage of a video pumping unit. This method mainly includes: first, extracting frames from a given video of a pumping unit to prepare for target detection of the pumping unit in the images; labeling the dataset of the pumping unit to train a Yolov5 target detection model; using the trained Yolov5 model to detect the pumping unit in the images, obtaining the pumping unit's bounding box, including the upper left corner (x1, y1) and lower right corner (x2, y2); in the current frame, cropping out the bounding box region of the pumping unit as image 2; locating the image T seconds before or after the current frame, cropping out the corresponding region in that frame as image 1; using a sliding window to take M×N windows on the two images, calculating the similarity between each pair of corresponding windows, and averaging to obtain the total similarity between the two images.

[0004] However, further research reveals that the aforementioned patent documents target oil pumping equipment for testing. To meet the specific requirement of testing oxygen cylinders and acetylene cylinders transported in the same vehicle, new efforts and attempts are still needed by those skilled in the art. Summary of the Invention

[0005] This invention provides a deep learning-based method for detecting the simultaneous transport of oxygen and acetylene cylinders in the same vehicle. This method can be used to detect violations such as transporting oxygen and acetylene cylinders in the same vehicle in real time. It can not only meet the management requirements of oilfield enterprises, but also promptly detect the aforementioned safety hazards and issue early warnings to prevent accidents from occurring.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0007] A deep learning-based method for detecting the simultaneous transport of oxygen cylinders and acetylene cylinders in the same vehicle includes the following steps:

[0008] Step S1: Obtain the monitoring video of the fire operation site, and use the group frame extraction method to obtain the image of the transport vehicle containing acetylene cylinders and oxygen cylinders from the monitoring video of the fire operation site;

[0009] Step S2: Annotate the image of the transport vehicle containing acetylene cylinders and oxygen cylinders obtained in Step S1, and generate sample data;

[0010] Step S3: Construct the YOLOv5 network model structure; train the YOLOv5 network model using the sample data obtained in step S2 to obtain a trained neural network model;

[0011] Step S4: Acquire real-time monitoring video of the work site, and acquire multiple consecutive frames of the site image to be tested at intervals as a group of test images;

[0012] Step S5: Input the image of the scene to be tested into the trained neural network model obtained in step S3 to obtain the detection target results of acetylene cylinder, oxygen cylinder and transport vehicle in the image of the scene to be tested;

[0013] Step S6: If both acetylene cylinder and oxygen cylinder detection targets are present in the image of the scene to be tested, further determine whether there is a transport vehicle detection target in the image of the scene to be tested.

[0014] If the detection target of the transport vehicle is present in the image of the site to be tested, calculate the overlapping coordinate sets between the detection target of the acetylene cylinder and the detection target of the transport vehicle, and the overlapping coordinate sets between the detection target of the oxygen cylinder and the detection target of the transport vehicle; and calculate the intersection-union ratio of the overlapping coordinate sets between the detection target of the acetylene cylinder and the detection target of the transport vehicle, and the overlapping coordinate sets between the detection target of the oxygen cylinder and the detection target of the transport vehicle, so as to determine whether the acetylene cylinder and the oxygen cylinder are on the same transport vehicle;

[0015] Step S7: Calculate the coordinate distance between the detection target results of the acetylene cylinder and the detection target results of the oxygen cylinder.

[0016] Preferably, the process of annotating the image of the transport vehicle containing acetylene cylinders and oxygen cylinders obtained in step S1 in step S2 can be specifically described as follows: the acetylene cylinders, oxygen cylinders, and transport vehicle in the image of the transport vehicle containing acetylene cylinders and oxygen cylinders obtained in step S1 are annotated separately.

[0017] Preferably, 75% of the sample data generated in step S2 is used as training samples, and 25% of the sample data generated in step S2 is used as test samples.

[0018] Preferably, in step S3, when constructing the YOLOv5 network model structure, the sampling feature maps are used at 8x and / or 16x.

[0019] Preferably, in step S4, the number of consecutive frames of the test site image acquired at intervals is not less than 45 frames.

[0020] Preferably, the detection target result of the oxygen cylinder in step 6 is defined as O2, and the coordinates of the detection target result of the oxygen cylinder are defined as (O2, O ... x1 O y1 ;O x2 O y2 The detection target result of the acetylene cylinder in step 6 is defined as CH; the coordinates of the detection target result of the acetylene cylinder are defined as (CH). x1 CH y1 ;CH x2 CH y2 The detection target result of the transport vehicle in step 6 is defined as trans; the coordinates of the detection target result of the transport vehicle are defined as (T x1 ,T y1 ;T x2 ,T y2 );

[0021] The formula for calculating the intersection-union ratio (IOU) between the overlapping coordinate sets of the detection target results of the acetylene cylinder and the transport vehicle, and the overlapping coordinate sets of the detection target results of the oxygen cylinder and the transport vehicle, satisfies: IOU = C⋃G / C⋂G; where C represents the overlapping coordinate set of the detection target results of the acetylene cylinder and the transport vehicle, and G represents the overlapping coordinate set of the detection target results of the oxygen cylinder and the transport vehicle.

[0022] In step S7, the coordinate distance between the detection target results of the acetylene cylinder and the oxygen cylinder satisfies: Distance = MIN|Ox{ 1，2} - CHx{ 1，2}|.

[0023] Preferably, after determining that the detection target of the transport vehicle exists in the image of the scene to be tested, and before comparing the coordinate positions between the detection target of the acetylene cylinder and the detection target of the transport vehicle, and between the detection target of the oxygen cylinder and the detection target of the transport vehicle, step S6 further includes: performing color enhancement processing on the image of the scene to be tested from three aspects: brightness, hue, and saturation, so as to increase the robustness of the image of the scene to be tested.

[0024] A more preferred embodiment also includes step S8: when the coordinate distance between the detection target result of the acetylene cylinder and the detection target result of the oxygen cylinder is less than a preset value, a trigger warning is issued;

[0025] The preset value satisfies: preset value = |O x1 -O x2 |*n, where n = (length of the transport vehicle) / (length of the oxygen cylinder + length of the acetylene cylinder).

[0026] This invention provides a deep learning-based method for detecting the simultaneous transport of oxygen and acetylene cylinders in the same vehicle. The method includes steps such as constructing a YOLOv5 network model structure, training the neural network model, comparing the coordinate positions of the detection targets for the acetylene cylinder and the transport vehicle, and comparing the coordinate positions of the detection targets for the oxygen cylinder and the transport vehicle; determining whether the acetylene and oxygen cylinders are on the same transport vehicle based on the intersection-union ratio (IUU); and calculating the coordinate distance between the detection targets for the acetylene and oxygen cylinders. This deep learning-based method for detecting the simultaneous transport of oxygen and acetylene cylinders, with the aforementioned steps, can be used to achieve real-time detection of violations such as the simultaneous transport of oxygen and acetylene cylinders in the same vehicle. It not only meets the management requirements of oilfield enterprises but also enables timely detection of such safety hazards and the issuance of early warnings, preventing accidents from occurring. Attached Figure Description

[0027] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the following drawings:

[0028] Figure 1 The present invention provides a deep learning-based method for detecting unworn portable gas detectors.

[0029] Figure 2 One of the schematic diagrams of sample data provided by the present invention;

[0030] Figure 3 This is the second schematic diagram of the sample data provided by the present invention. Detailed Implementation

[0031] This invention provides a deep learning-based method for detecting the simultaneous transport of oxygen and acetylene cylinders in the same vehicle. This method can be used to detect violations such as transporting oxygen and acetylene cylinders in the same vehicle in real time. It can not only meet the management requirements of oilfield enterprises, but also promptly detect the aforementioned safety hazards and issue early warnings to prevent accidents from occurring.

[0032] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0033] Example 1

[0034] This invention provides a deep learning-based method for detecting the simultaneous transport of oxygen and acetylene cylinders in the same vehicle. Figure 1 As shown, it includes the following steps:

[0035] Step S1: Obtain the monitoring video of the fire operation site, and use the group frame extraction method to obtain the image of the transport vehicle containing acetylene cylinders and oxygen cylinders from the monitoring video of the fire operation site;

[0036] Specifically, monitoring videos of open flame operations can be obtained through cameras at the oilfield production site; alternatively, monitoring video data of open flame operations stored in the central control center can be referenced.

[0037] Step S2: Annotate the image of the transport vehicle containing acetylene cylinders and oxygen cylinders obtained in Step S1, and generate sample data;

[0038] In a preferred embodiment of the present invention, the process of annotating the image of the transport vehicle containing acetylene cylinders and oxygen cylinders obtained in step S1 in step S2 can be specifically described as follows: the acetylene cylinders, oxygen cylinders, and transport vehicle in the image of the transport vehicle containing acetylene cylinders and oxygen cylinders obtained in step S1 are annotated separately. Furthermore, more preferably, in the generated sample data, 75% is selected as training samples, and the remaining 25% is used as test samples.

[0039] Step S3: Construct the YOLOv5 network model structure; train the YOLOv5 network model using the sample data obtained in step S2 to obtain a trained neural network model;

[0040] It should be noted that when constructing the YOLOv5 network model structure, it is preferable to use sampling feature maps at 8x and / or 16x. For specific examples of these sampling feature maps, please refer to [link / reference needed]. Figure 2 , Figure 3As shown; and during the training of the YOLOv5 network model, according to the preferred sample data classification method in step S2 above, 75% of the sample data is used as training samples and the remaining 25% is used as test samples for training, so as to obtain a trained neural network model.

[0041] Step S4: Acquire real-time monitoring video of the work site, and acquire multiple consecutive frames of the site image to be tested at intervals as a group of test images;

[0042] In a preferred embodiment of the present invention, the number of image frames acquired in step S4 in a series of consecutive test site images is not less than 45 frames.

[0043] Step S5: Input the image of the scene to be tested into the trained neural network model obtained in step S3 to obtain the detection target results of acetylene cylinder, oxygen cylinder and transport vehicle in the image of the scene to be tested;

[0044] Step S6: If both acetylene cylinder and oxygen cylinder detection targets are present in the image of the scene to be tested, further determine whether there is a transport vehicle detection target in the image of the scene to be tested.

[0045] If the detection target of the transport vehicle is present in the image of the site to be tested, the overlapping coordinate sets between the detection target of the acetylene cylinder and the detection target of the transport vehicle, and the overlapping coordinate sets between the detection target of the oxygen cylinder and the detection target of the transport vehicle are calculated respectively; and the intersection-union ratio of the overlapping coordinate sets between the detection target of the acetylene cylinder and the detection target of the transport vehicle, and the overlapping coordinate sets between the detection target of the oxygen cylinder and the detection target of the transport vehicle are calculated to determine whether the acetylene cylinder and the oxygen cylinder are on the same transport vehicle;

[0046] Step S7: Calculate the coordinate distance between the detection target results of the acetylene cylinder and the detection target results of the oxygen cylinder.

[0047] At this point, the detection of the test group (images) consisting of multiple frames of the test site images has been completed.

[0048] Example 2

[0049] Example 2 includes all the technical features of Example 1; the present invention provides a deep learning-based method for detecting the simultaneous transport of oxygen cylinders and acetylene cylinders in the same vehicle, such as... Figure 1 As shown, it includes the following steps:

[0050] Step S1: Obtain the monitoring video of the fire operation site, and use the group frame extraction method to obtain the image of the transport vehicle containing acetylene cylinders and oxygen cylinders from the monitoring video of the fire operation site;

[0051] Specifically, monitoring videos of open flame operations can be obtained through cameras at the oilfield production site; alternatively, monitoring video data of open flame operations stored in the central control center can be referenced.

[0052] Step S2: Annotate the image of the transport vehicle containing acetylene cylinders and oxygen cylinders obtained in Step S1, and generate sample data;

[0053] In a preferred embodiment of the present invention, the process of annotating the image of the transport vehicle containing acetylene cylinders and oxygen cylinders obtained in step S1 in step S2 can be specifically described as follows: the acetylene cylinders, oxygen cylinders, and transport vehicle in the image of the transport vehicle containing acetylene cylinders and oxygen cylinders obtained in step S1 are annotated separately. Furthermore, more preferably, in the generated sample data, 75% is selected as training samples, and the remaining 25% is used as test samples.

[0054] Step S3: Construct the YOLOv5 network model structure; train the YOLOv5 network model using the sample data obtained in step S2 to obtain a trained neural network model;

[0055] It should be noted that when constructing the YOLOv5 network model structure, it is preferable to use sampling feature maps at 8x and / or 16x. For specific examples of these sampling feature maps, please refer to [link / reference needed]. Figure 2 As shown; and during the training of the YOLOv5 network model, according to the preferred sample data classification method in step S2 above, 75% of the sample data is used as training samples and the remaining 25% is used as test samples for training, so as to obtain a trained neural network model.

[0056] Step S4: Acquire real-time monitoring video of the work site, and acquire multiple consecutive frames of the site image to be tested at intervals as a group of test images;

[0057] In a preferred embodiment of the present invention, the number of image frames acquired in step S4 in a series of consecutive test site images is not less than 45 frames.

[0058] Step S5: Input the image of the scene to be tested into the trained neural network model obtained in step S3 to obtain the detection target results of acetylene cylinder, oxygen cylinder and transport vehicle in the image of the scene to be tested;

[0059] Step S6: If both acetylene cylinder and oxygen cylinder detection targets are present in the image of the scene to be tested, further determine whether there is a transport vehicle detection target in the image of the scene to be tested.

[0060] If the detection target of the transport vehicle is present in the image of the site to be tested, the overlapping coordinate sets between the detection target of the acetylene cylinder and the detection target of the transport vehicle, and the overlapping coordinate sets between the detection target of the oxygen cylinder and the detection target of the transport vehicle are calculated respectively; and the intersection-union ratio of the overlapping coordinate sets between the detection target of the acetylene cylinder and the detection target of the transport vehicle, and the overlapping coordinate sets between the detection target of the oxygen cylinder and the detection target of the transport vehicle are calculated to determine whether the acetylene cylinder and the oxygen cylinder are on the same transport vehicle;

[0061] Step S7: Calculate the coordinate distance between the detection target results of the acetylene cylinder and the detection target results of the oxygen cylinder.

[0062] At this point, the detection of the test group (images) consisting of multiple frames of the test site images has been completed.

[0063] Furthermore, Example 2 further defines the detection target results of oxygen cylinders, the detection target results of acetylene cylinders, and the horizontal coordinates of the detection target of acetylene cylinders as follows.

[0064] Specifically, the detection target result of the oxygen cylinder in step 6 is defined as O2, and the coordinates of the detection target result of the oxygen cylinder are defined as (O2, O ... x1 O y1 ;O x2 O y2 The detection target result of the acetylene cylinder in step 6 is defined as CH; the coordinates of the detection target result of the acetylene cylinder are defined as (CH). x1 CH y1 ;CH x2 CH y2 The detection target result of the transport vehicle in step 6 is defined as trans; the coordinates of the detection target result of the transport vehicle are defined as (T x1 ,T y1 ;T x2 ,T y2 );

[0065] The formula for calculating the intersection-union ratio (IOU) between the overlapping coordinate sets of the detection target results of the acetylene cylinder and the transport vehicle, and the overlapping coordinate sets of the detection target results of the oxygen cylinder and the transport vehicle, satisfies: IOU = C⋃G / C⋂G; where C represents the overlapping coordinate set of the detection target results of the acetylene cylinder and the transport vehicle, and G represents the overlapping coordinate set of the detection target results of the oxygen cylinder and the transport vehicle.

[0066] In step S7, the coordinate distance between the detection target results of the acetylene cylinder and the oxygen cylinder satisfies: Distance = MIN|Ox{ 1，2} - CHx{ 1，2}|.

[0067] The coordinate distance between the detection target results of the acetylene cylinder and the oxygen cylinder is calculated and compared with the preset value. When the coordinate distance between the detection target results of the acetylene cylinder and the oxygen cylinder is less than the preset value, the acetylene cylinder and the oxygen cylinder are on the same transport vehicle, thereby triggering the linkage warning in step S8.

[0068] It is worth noting that this preset value is related to the size of the transport vehicle and the sizes of the acetylene and oxygen cylinders; different vehicles transporting acetylene and oxygen cylinders on-site will have different cylinder sizes, and therefore the preset value will also differ. Specifically, the preset value satisfies: Preset value = |O x1 -O x2 |*n, where n = (length of the transport vehicle) / (length of the oxygen cylinder + length of the acetylene cylinder).

[0069] Example 3

[0070] Embodiment 3 includes all the technical features of Embodiments 1 and 2, and due to space limitations, they will not be elaborated upon here.

[0071] Furthermore, this embodiment three further includes the following: after determining that the detection target result of the transport vehicle exists in the image of the scene to be tested, and before comparing the coordinate positions between the detection target result of the acetylene cylinder and the detection target result of the transport vehicle, and between the coordinate positions between the detection target result of the oxygen cylinder and the detection target result of the transport vehicle, step S6 further includes: performing color enhancement processing on the image of the scene to be tested from three aspects: brightness, hue, and saturation, so as to increase the robustness of the image of the scene to be tested.

[0072] It should be added that color enhancement processing can improve the robustness of the test site image, simulate the test site image under different weather conditions, and optimize the occlusion problem that may occur in the real-time video where the test site image is located (using random erasure to simulate target occlusion, so that occluded objects can be quickly identified during real-time monitoring), ultimately achieving a significant improvement in detection performance.

[0073] Of course, the above implementation is for reference only and is not intended to further limit the present invention.

[0074] This invention provides a deep learning-based method for detecting the simultaneous transport of oxygen and acetylene cylinders in the same vehicle. The method includes steps such as constructing a YOLOv5 network model structure, training the neural network model, comparing the coordinate positions of the detection targets for the acetylene cylinder and the transport vehicle, and comparing the coordinate positions of the detection targets for the oxygen cylinder and the transport vehicle; determining whether the acetylene and oxygen cylinders are on the same transport vehicle based on the intersection-union ratio (IUU); and calculating the coordinate distance between the detection targets for the acetylene and oxygen cylinders. This deep learning-based method for detecting the simultaneous transport of oxygen and acetylene cylinders, with the aforementioned steps, can be used to achieve real-time detection of violations such as the simultaneous transport of oxygen and acetylene cylinders in the same vehicle. It not only meets the management requirements of oilfield enterprises but also enables timely detection of such safety hazards and the issuance of early warnings, preventing accidents from occurring.

[0075] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for detecting the simultaneous transport of oxygen cylinders and acetylene cylinders in the same vehicle based on deep learning, characterized in that, The steps include the following: Step S1: Obtain the monitoring video of the fire operation site, and use the group frame extraction method to obtain the image of the transport vehicle containing acetylene cylinders and oxygen cylinders from the monitoring video of the fire operation site; Step S2: Annotate the image of the transport vehicle containing acetylene cylinders and oxygen cylinders obtained in Step S1, and generate sample data; Step S3: Construct the YOLOv5 network model structure; train the YOLOv5 network model using the sample data obtained in step S2 to obtain a trained neural network model; Step S4: Acquire real-time monitoring video of the work site, and acquire multiple consecutive frames of the site image to be tested at intervals as a group of test images; Step S5: Input the image of the scene to be tested into the trained neural network model obtained in step S3 to obtain the detection target results of acetylene cylinder, oxygen cylinder and transport vehicle in the image of the scene to be tested; Step S6: If both acetylene cylinder and oxygen cylinder detection targets are present in the image of the scene to be tested, further determine whether there is a transport vehicle detection target in the image of the scene to be tested. If the detection target result of the transport vehicle exists in the image of the site to be tested, then calculate the overlapping coordinate set between the detection target result of the acetylene cylinder and the detection target result of the transport vehicle, and the overlapping coordinate set between the detection target result of the oxygen cylinder and the detection target result of the transport vehicle, respectively. The intersection-union ratio of the overlapping coordinate sets between the detection target results of the acetylene cylinder and the detection target results of the transport vehicle, and the overlapping coordinate sets between the detection target results of the oxygen cylinder and the detection target results of the transport vehicle, is calculated to determine whether the acetylene cylinder and the oxygen cylinder are on the same transport vehicle. Step S7: Calculate the coordinate distance between the detection target results of the acetylene cylinder and the detection target results of the oxygen cylinder; It also includes step S8: when the coordinate distance between the detection target result of the acetylene cylinder and the detection target result of the oxygen cylinder is less than a preset value, a trigger warning is issued; The preset value satisfies: Preset value = Where n = (length of transport vehicle) / (length of oxygen cylinder + length of acetylene cylinder).

2. The method for detecting the simultaneous transport of oxygen cylinders and acetylene cylinders in the same vehicle based on deep learning, as described in claim 1, is characterized in that... The process of annotating the image of the transport vehicle containing acetylene cylinders and oxygen cylinders obtained in step S1 in step S2 is specifically described as follows: the acetylene cylinders, oxygen cylinders, and transport vehicle in the image of the transport vehicle containing acetylene cylinders and oxygen cylinders obtained in step S1 are annotated separately.

3. The method for detecting the simultaneous transport of oxygen cylinders and acetylene cylinders in the same vehicle based on deep learning, as described in claim 1, is characterized in that... 75% of the sample data generated in step S2 is used as training samples, and 25% of the sample data generated in step S2 is used as test samples.

4. The method for detecting the simultaneous transport of oxygen cylinders and acetylene cylinders in the same vehicle based on deep learning, as described in claim 1, is characterized in that... In step S3, during the construction of the YOLOv5 network model structure, feature maps are sampled at 8x and / or 16x.

5. The method for detecting the simultaneous transport of oxygen cylinders and acetylene cylinders in the same vehicle based on deep learning, as described in claim 1, is characterized in that... In step S4, the number of consecutive frames of images of the test site acquired at intervals shall not be less than 45 frames.

6. The method for detecting the simultaneous transport of oxygen cylinders and acetylene cylinders in the same vehicle based on deep learning, as described in claim 1, is characterized in that... In step 6, the target result for the oxygen cylinder detection is defined as O2, and the coordinates of the target result for the oxygen cylinder detection are defined as (O... x1 O y1 ;O x2 O y2 The detection target result of the acetylene cylinder in step 6 is defined as CH; the coordinates of the detection target result of the acetylene cylinder are defined as (CH). x1 CH y1 ;CH x2 CH y2 The detection target result of the transport vehicle in step 6 is defined as trans; the coordinates of the detection target result of the transport vehicle are defined as (T x1 ,T y1 ;T x2 ,T y2 ); The formula for calculating the intersection-union ratio (IOU) between the overlapping coordinate sets of the detection target results of the acetylene cylinder and the transport vehicle, and the overlapping coordinate sets of the detection target results of the oxygen cylinder and the transport vehicle, satisfies: IOU=(C∩G) / (C∪G); where C refers to the overlapping coordinate set between the detection target results of the acetylene cylinder and the transport vehicle, and G refers to the overlapping coordinate set between the detection target results of the oxygen cylinder and the transport vehicle. In step S7, the coordinate distance between the detection target results of the acetylene cylinder and the oxygen cylinder satisfies: Distance = MIN | Ox{ 1，2 }-CHx{ 1，2 }|.

7. The method for detecting the simultaneous transport of oxygen cylinders and acetylene cylinders in the same vehicle based on deep learning, as described in claim 1, is characterized in that... Step S6, after determining that the detection target of the transport vehicle exists in the image of the scene to be tested, and before comparing the coordinate positions between the detection target of the acetylene cylinder and the detection target of the transport vehicle, and between the detection target of the oxygen cylinder and the detection target of the transport vehicle, also includes: performing color enhancement processing on the image of the scene to be tested from three aspects: brightness, hue, and saturation, in order to increase the robustness of the image of the scene to be tested.

Images