Stack Exchange Network

Stack Exchange network consists of 174 Q&A communities including Stack Overflow, the largest, most trusted online community for developers to learn, share their knowledge, and build their careers.

Visit Stack Exchange

Questions tagged [object-detection]

The tag has no usage guidance.

0
votes
1answer
20 views

mean average precision - pseudo code

There are many blog posts and threads concerning the computation of mAP, both for PASCAL and for COCO. Still, they do not answer some fundamental questions. I am listing here the flow as I understand ...
0
votes
0answers
12 views

Using single cell image to train model to detect and recognize multi cell image; different size of target cell in training and test set

I have a dataset containing images, each image has a single cell along with some smaller inconsequential cells in the background. Can this dataset be used to train a model to detect and recognize ...
0
votes
1answer
23 views

Mask RCNN: Random predictions during inference for the same image

I recently trained the Mask RCNN (matterport's implementation) on some satellite images, but during inference mode, I'm getting random predictions for the same set of weights for the same image. That ...
1
vote
0answers
14 views

Keras - How to using Cnn trained model on video image.?

I have trained Cat and dog image using CNN with Python ( total 10K, 8K for training and 2K for testing ). I want to make prediction in a video. Video Contain both cat and dog at a single moment. How ...
0
votes
0answers
12 views

How is Stochastic Gradient Descent done in Faster RCNN?

In Tenosrflow Object Detection API, you can choose different pretrained models such as Faster RCNN, SSD. Then you can specify the batch size during training. I know that for Stochastic Gradient ...
0
votes
0answers
12 views

How much CuDnn has impact on DNN performance in object detection?

I am not sure if I am in the right place to ask this question. But I am really confused. I am using JetsonTX1 for running DNN model(Yolov3) for real-time object detection. When the CuDNN option is ...
0
votes
0answers
26 views

Fluctuating Learning rate in tensorboard

I am trying to train an faster rcnn for object detection. While I am trying to fine-tune the object detection network I am seeing my learning rate is fluctuating even though I didn't change it. The ...
1
vote
0answers
24 views

Using tensorflow object detection in another model

I am trying to use Tensorflow (tf) object detection API models in another custom model I built. Specifically, I am trying to do: jointly train tf object detection models Y with another model X. in a ...
0
votes
2answers
24 views

Tattoo Image Recognition - Should I Crop Training Data Background

I am trying to train a neural network to detect objects within a tattoo. I couldn't find any existing labeled dataset so I need to manually create and label my own. I only understand the basics of ...
0
votes
0answers
11 views

Understanding YOLO Loss Function

In the following formulation, if xi and yi represent center of anchor box, shouldnt we have ...
2
votes
1answer
37 views

Very slow convergence with CNN [closed]

I am new to deep learning. I am working on training an SSD model on a set of small objects. I am using Adam gradient descent for optimization and a large input (800x800), but I seem to only get an ...
0
votes
0answers
7 views

Literature Resources on Object Detection

I would like to learn as much as I can about object detection, including the history and math behind it. My plan is to build up my knowledge all the way to state of the art object detection techniques ...
0
votes
1answer
27 views

How to increase accuracy of model from tensorflow model zoo?

Situation: My dataset is 70k images of people wearing clothes. Images are labeled: bbox position and class. There are 10 classes. I did 80:20 split. Categories are balanced with exception of one ...
1
vote
1answer
26 views

The mix of leaky Relu at the first layers of CNN along with conventional Relu for object detection

First of all, I know the usage of leaky RELUs and some other relevant leaky activation functions as well. However I have seen in a lot of papers on object detection tasks (e.g YOLO) to use this type ...
1
vote
0answers
17 views

Training detector without bounding box data

From what I can see most object detection NNs (Fast(er) R-CNN, YOLO etc) are trained on data including bounding boxes indicating where in the picture the objects are localized. Is there any model ...
-1
votes
0answers
43 views

Tensorflow object detection API issues

I use Tensorflow object detection API, and have 2 questions: Where can i find all data augmentation parameters that available for config file? When I train model on my own dataset, i see loss metric ...
0
votes
1answer
8 views

How to obtain and load a good initial data set for object localization?

I'm looking for a good data set for training a CNN based network to do object localization (i.e. a data set with class labels and bounding box data). What is a good initial data set to use? How can ...
0
votes
1answer
11 views

How to train object detection system for 2 classes having two seperate datasets for each class?

I have dataset of class A and a dataset of class B. However dataset A does not contain annotated class B and vice-versa. Is there a way to somehow train object detection system like SSD to detect ...
1
vote
0answers
17 views

Does anyone have suggestions for how Goodfellow et al detected digits from the larger Street View images in their 2014 paper?

In their 2014 paper titled, "Multi-digit Number Recognition from Street View Imagery using Deep Convolutional Neural Networks", they write, We use an automated method (beyond the scope of this ...
-1
votes
0answers
44 views

Framework to build object detection model to predict identical or closely identical images

I am looking for an approach to build Image classification and Object detection model for a large category of items (About 200K) but with the very smaller number(4 images on each) of subject input on ...
0
votes
0answers
12 views

Are there any algorithms for human detection/tracking that work from all angles?

I've looked at Haar/HOG in OpenCV for cheap human detection/tracking, but it's completely unclear if it works from angles other than side-on. I've been looking at some deep learning solutions, but it'...
1
vote
1answer
60 views

Get bounding boxes for adjacent instances of a single class in image

I have a dataset with thousands of music score pages and manually annotated bounding boxes for the individual bars: My objective is now to train a DNN that should ultimately be able to get these ...
1
vote
1answer
73 views

Bounding Boxes in YOLO Model

The YOLO model splits the image into smaller boxes and each box is responsible for predicting 5 bounding boxes. My question is how does the model make these bounding boxes for every grid cell ? Does ...
0
votes
1answer
116 views

Tensorflow: Unable to Save checkpoint after every 2 global steps during training the SSD model object detection

python models/object_detection/train.py --logtostderr --train_dir=training/ --pipeline_config_path=training/ssd_inception_v2_coco.config INFO:tensorflow:Restoring parameters from /home/rahul/...
1
vote
0answers
112 views

SSD based on ResNet-101 doesn't improve over SSD-VGGNet

I am training a SSD model for detecting mobile cranes. The training dataset contains 1000 images and test set over 400 images. About 200 epochs gave mAP 83%, but my target is 90%. So I trained SSD-...
1
vote
0answers
14 views

What does AR(max=1) mean?

http://cocodataset.org/#detection-eval "ARmax=1: AR given 1 detection per image" "AR is the maximum recall given a fixed number of detections per image, averaged over categories and IoUs." Above link,...
1
vote
0answers
27 views

Add training data to YOLO post-training

I've been playing around with YOLOv3 and obtaining some good results on the ~20 custom classes I trained. However, one or two classes look like they can use some additional training data (not a lot, ...
4
votes
1answer
117 views

What is difference between intersection over union (IoU) and intersection over bounding box (IoBB)?

Can someone give a detailed explanation IoU and IoBB along with that the differences between them.
1
vote
0answers
41 views

Training a convoltion neural network for localization

I am new into convolutional neural networks and I am trying to build a cnn for localization of object on tensorflow. Currently I am trying to replicate a cnn similar to that of Alexnet and then want ...
2
votes
2answers
77 views

Do you apply confidence threshold before calculating mAP for YOLOv2?

I've been using YOLO v2 for object detection and have been trying to use mAP for validating the model. For training I use a confidence threshold of 0.25, however I am unsure as to whether to apply ...
2
votes
0answers
35 views

Better to crop or compress training data?

I've been trying to make an object recogniser in Tensorflow and have used labelImg to classify large electrical transmission towers at varying distances. In order to make 10-16MP (~2-7MB) images train ...
0
votes
0answers
25 views

AVOD Object Detection Library Using Point Cloud and Image Data

I downloaded and used the AVOD object detection library using the code and instructions available on the following github link. https://github.com/kujason/avod I ...
1
vote
0answers
18 views

Detecting directions using Convolutional neural networks [closed]

I am working on a task where I have to detect damages on the vehicles and exactly where the damage has occurred. So I have to not only detect the damage on the door but also mentioned which door(front ...
3
votes
1answer
68 views

How to label “other” while labeling image for object detection/classification?

I want to train a model to recognize the different categories of food e.g. rice, burger, apple, pizza, orange and other things. After the first training, I realized that the model is detecting other ...
0
votes
0answers
141 views

How to calculate distance between objects Yolov2?

I want to calculate distance between every objects in yolo v2 . how would i able to perform this ? I am new to yolo v2 and after some fact finding i would not be able to find something on the internet
1
vote
1answer
26 views

Detecting person on the picture if we know it is there

I implemented an algorithm for pedestrian classification using Keras and Tensorflow and it is quite good, it has 96 percent accuracy on test data. It gives you information about existence of a person ...
5
votes
1answer
851 views

Does image's background matter for detector training (CNN)?

Does an image's background matter for detector/localisation in the training part (using CNN)? For example, if I want to make a face detector, which one is better as training dataset? Faces cropped ...
4
votes
1answer
388 views

What is the difference between tensorflow saved_model.pb and frozen_inference_graph.pb?

I've re-trained a model (following this tutorial) from the google's object detection zoo (ssd_inception_v2_coco) on a WIDER Faces Dataset and it seems to work if I use ...
2
votes
1answer
74 views

Object Detection classification

I am currently training a classifier for detecting resistors using TensorFlow Object Detection API. For that, I downloaded resistor images from ImageNet and I am currently labeling those who will be ...
0
votes
0answers
25 views

Most simple algorithm for pedestrian detection

I would like to implement a simple pedestrian detection algorithm. I would like to know which one is most simple to implement. I want to use neural networks and Keras. The algorithm should predict if ...
0
votes
0answers
181 views

Retraining an object detection model on fewer classes

I have been trying to create an object detection model with tensorflow supposed to detect only 2 classes of objects on an embedded device. I've first tried with ssd_mobilenet_v2_coco, it works but ...
1
vote
1answer
591 views

How to download images and bounding boxes from imageNet?

I am doing object detection for a specific class, say, chairs . I want to download images of chairs from imageNet. I also want to download the annotation xml files ...
0
votes
2answers
159 views

How to train model on specific classes from dataset for object detection?

I want to use darkflow to perform object detection. I have 5 classes of objects I want to detect, say , ...
0
votes
0answers
48 views

customizing haar cascade classifier

I have implemented the haar cascade classification for object detection. I want to extract daubechies features in the same way and apply a classifier of my interest. How to customize haar cascade ...
2
votes
0answers
22 views

Contextual Object Detection

An example of what I'd like to do is identify the price of a product on a product page. While I can train a CNN to identify prices, it's likely that it would recognise every instance of a price on a ...
3
votes
0answers
443 views

Preparing ground truth labels for YOLO3

I want to train YOLO3 for a custom dataset that has raw labels in JSON format. Each bounding box in JSON is specified as [x1, y1, x2, y2]. So far, I have ...
3
votes
1answer
2k views

mAP scores on tensorboard (Tensorflow Object Detection API) are all 0 even though the loss value is low

I trained a faster-rcnn model on the tensorflow object detection API on a custom dataset. I found that the loss is ~2 after 3.5k steps. However, when I ran eval.py, the mAP scores are all almost 0 as ...
1
vote
1answer
51 views

How to setup my Ubuntu PC for these Deep Learning frameworks? [closed]

I need to use both NVIDIA Digits and tensorflow Object Detection API for different deep learning purposes. I am a bit lost with the compatibility issues, especially for what concerns CUDA. So which ...
0
votes
1answer
114 views

Space between an object and the ground truth bounding box

Which way is better for drawing the ground truth boxes for object detection? Drawing as tight as the sides of the box touch the border of the object, or Make a little space between the box and the ...
9
votes
5answers
2k views

Unsupervised image segmentation

I am trying to implement an algorithm where given an image with several objects on a plane table, desired is the output of segmentation masks for each object. Unlike in CNN's, the objective here is to ...