本代码理论上可以通杀市面上所有滑块(如果真出现了识别不到,请自行根据文章末尾的参考内容,自己进行调教)
在线测试地址:simple_ocr
截至.7.13,可以破解的有:
1、顶象
2、网易易盾
以下是识别代码:
from io import BytesIOimport onnxruntimeimport torchimport torchvisionimport numpy as npfrom mon.by import Byfrom selenium.webdriver.chrome.options import Optionsfrom selenium import webdriverimport timeimport cv2import base64from selenium.webdriver import ActionChainsimport randomfrom PIL import Imagefrom lxml import etreeimport requestsdef padded_resize(im, new_shape=(640, 640), stride=32):try:shape = im.shape[:2]r = min(new_shape[0] / shape[0], new_shape[1] / shape[1])new_unpad = int(round(shape[1] * r)), int(round(shape[0] * r))dw, dh = new_shape[1] - new_unpad[0], new_shape[0] - new_unpad[1]# dw, dh = np.mod(dw, stride), np.mod(dh, stride)dw /= 2dh /= 2if shape[::-1] != new_unpad: # resizeim = cv2.resize(im, new_unpad, interpolation=cv2.INTER_LINEAR)top, bottom = int(round(dh - 0.1)), int(round(dh + 0.1))left, right = int(round(dw - 0.1)), int(round(dw + 0.1))im = cv2.copyMakeBorder(im, top, bottom, left, right, cv2.BORDER_CONSTANT, value=(114, 114, 114)) # add border# Convertim = im.transpose((2, 0, 1))[::-1] # HWC to CHW, BGR to RGBim = np.ascontiguousarray(im)im = torch.from_numpy(im)im = im.float()im /= 255im = im[None]im = im.cpu().numpy() # torch to numpyreturn imexcept:print("123")def xywh2xyxy(x):# Convert nx4 boxes from [x, y, w, h] to [x1, y1, x2, y2] where xy1=top-left, xy2=bottom-righty = x.clone() if isinstance(x, torch.Tensor) else np.copy(x)y[:, 0] = x[:, 0] - x[:, 2] / 2 # top left xy[:, 1] = x[:, 1] - x[:, 3] / 2 # top left yy[:, 2] = x[:, 0] + x[:, 2] / 2 # bottom right xy[:, 3] = x[:, 1] + x[:, 3] / 2 # bottom right yreturn ydef box_iou(box1, box2):"""Return intersection-over-union (Jaccard index) of boxes.Both sets of boxes are expected to be in (x1, y1, x2, y2) format.Arguments:box1 (Tensor[N, 4])box2 (Tensor[M, 4])Returns:iou (Tensor[N, M]): the NxM matrix containing the pairwiseIoU values for every element in boxes1 and boxes2"""def box_area(box):# box = 4xnreturn (box[2] - box[0]) * (box[3] - box[1])area1 = box_area(box1.T)area2 = box_area(box2.T)# inter(N,M) = (rb(N,M,2) - lt(N,M,2)).clamp(0).prod(2)inter = (torch.min(box1[:, None, 2:], box2[:, 2:]) - torch.max(box1[:, None, :2], box2[:, :2])).clamp(0).prod(2)return inter / (area1[:, None] + area2 - inter) # iou = inter / (area1 + area2 - inter)def non_max_suppression(prediction, conf_thres=0.25, iou_thres=0.45, classes=None, agnostic=False, multi_label=False,labels=(), max_det=300):"""Runs Non-Maximum Suppression (NMS) on inference resultsReturns:list of detections, on (n,6) tensor per image [xyxy, conf, cls]"""nc = prediction.shape[2] - 5 # number of classesxc = prediction[..., 4] > conf_thres # candidates# Checksassert 0 <= conf_thres <= 1, f'Invalid Confidence threshold {conf_thres}, valid values are between 0.0 and 1.0'assert 0 <= iou_thres <= 1, f'Invalid IoU {iou_thres}, valid values are between 0.0 and 1.0'# Settingsmin_wh, max_wh = 2, 7680 # (pixels) minimum and maximum box width and heightmax_nms = 30000 # maximum number of boxes into torchvision.ops.nms()time_limit = 10.0 # seconds to quit afterredundant = True # require redundant detectionsmulti_label &= nc > 1 # multiple labels per box (adds 0.5ms/img)merge = False # use merge-NMSt = time.time()output = [torch.zeros((0, 6), device=prediction.device)] * prediction.shape[0]for xi, x in enumerate(prediction): # image index, image inference# Apply constraintsx[((x[..., 2:4] < min_wh) | (x[..., 2:4] > max_wh)).any(1), 4] = 0 # width-heightx = x[xc[xi]] # confidence# Cat apriori labels if autolabellingif labels and len(labels[xi]):lb = labels[xi]v = torch.zeros((len(lb), nc + 5), device=x.device)v[:, :4] = lb[:, 1:5] # boxv[:, 4] = 1.0 # confv[range(len(lb)), lb[:, 0].long() + 5] = 1.0 # clsx = torch.cat((x, v), 0)# If none remain process next imageif not x.shape[0]:continue# Compute confx[:, 5:] *= x[:, 4:5] # conf = obj_conf * cls_conf# Box (center x, center y, width, height) to (x1, y1, x2, y2)box = xywh2xyxy(x[:, :4])# Detections matrix nx6 (xyxy, conf, cls)if multi_label:i, j = (x[:, 5:] > conf_thres).nonzero(as_tuple=False).Tx = torch.cat((box[i], x[i, j + 5, None], j[:, None].float()), 1)else: # best class onlyconf, j = x[:, 5:].max(1, keepdim=True)x = torch.cat((box, conf, j.float()), 1)[conf.view(-1) > conf_thres]# Filter by classif classes is not None:x = x[(x[:, 5:6] == torch.tensor(classes, device=x.device)).any(1)]# Apply finite constraint# if not torch.isfinite(x).all():#x = x[torch.isfinite(x).all(1)]# Check shapen = x.shape[0] # number of boxesif not n: # no boxescontinueelif n > max_nms: # excess boxesx = x[x[:, 4].argsort(descending=True)[:max_nms]] # sort by confidence# Batched NMSc = x[:, 5:6] * (0 if agnostic else max_wh) # classesboxes, scores = x[:, :4] + c, x[:, 4] # boxes (offset by class), scoresi = torchvision.ops.nms(boxes, scores, iou_thres) # NMSif i.shape[0] > max_det: # limit detectionsi = i[:max_det]if merge and (1 < n < 3E3): # Merge NMS (boxes merged using weighted mean)# update boxes as boxes(i,4) = weights(i,n) * boxes(n,4)iou = box_iou(boxes[i], boxes) > iou_thres # iou matrixweights = iou * scores[None] # box weightsx[i, :4] = torch.mm(weights, x[:, :4]).float() / weights.sum(1, keepdim=True) # merged boxesif redundant:i = i[iou.sum(1) > 1] # require redundancyoutput[xi] = x[i]if (time.time() - t) > time_limit:break # time limit exceededreturn outputdef xyxy2xywh(x):# Convert nx4 boxes from [x1, y1, x2, y2] to [x, y, w, h] where xy1=top-left, xy2=bottom-righty = x.clone() if isinstance(x, torch.Tensor) else np.copy(x)y[:, 0] = (x[:, 0] + x[:, 2]) / 2 # x centery[:, 1] = (x[:, 1] + x[:, 3]) / 2 # y centery[:, 2] = x[:, 2] - x[:, 0] # widthy[:, 3] = x[:, 3] - x[:, 1] # heightreturn ydef is_ascii(s=''):# Is string composed of all ASCII (no UTF) characters? (note str().isascii() introduced in python 3.7)s = str(s) # convert list, tuple, None, etc. to strreturn len(s.encode().decode('ascii', 'ignore')) == len(s)def box_label(self, box, label='', color=(128, 128, 128), txt_color=(255, 255, 255)):# Add one xyxy box to image with labelif self.pil or not is_ascii(label):self.draw.rectangle(box, width=self.lw, outline=color) # boxif label:w, h = self.font.getsize(label) # text width, heightoutside = box[1] - h >= 0 # label fits outside boxself.draw.rectangle((box[0],box[1] - h if outside else box[1],box[0] + w + 1,box[1] + 1 if outside else box[1] + h + 1), fill=color)# self.draw.text((box[0], box[1]), label, fill=txt_color, font=self.font, anchor='ls') # for PIL>8.0self.draw.text((box[0], box[1] - h if outside else box[1]), label, fill=txt_color, font=self.font)else: # cv2p1, p2 = (int(box[0]), int(box[1])), (int(box[2]), int(box[3]))cv2.rectangle(self.im, p1, p2, color, thickness=self.lw, lineType=cv2.LINE_AA)if label:tf = max(self.lw - 1, 1) # font thicknessw, h = cv2.getTextSize(label, 0, fontScale=self.lw / 3, thickness=tf)[0] # text width, heightoutside = p1[1] - h - 3 >= 0 # label fits outside boxp2 = p1[0] + w, p1[1] - h - 3 if outside else p1[1] + h + 3cv2.rectangle(self.im, p1, p2, color, -1, cv2.LINE_AA) # filledcv2.putText(self.im, label, (p1[0], p1[1] - 2 if outside else p1[1] + h + 2), 0, self.lw / 3, txt_color,thickness=tf, lineType=cv2.LINE_AA)def return_coordinates(xyxy, conf):conf = float(conf.numpy())gain = 1.02pad = 10xyxy = torch.tensor(xyxy).view(-1, 4)b = xyxy2xywh(xyxy) # boxesb[:, 2:] = b[:, 2:] * gain + pad # box wh * gain + padxyxy = xywh2xyxy(b).long()c1, c2 = (int(xyxy[0, 0]) + 6, int(xyxy[0, 1]) + 6), (int(xyxy[0, 2]) - 6, int(xyxy[0, 3]) - 6)# print(f"leftTop:{c1},rightBottom:{c2},Confidence:{conf*100}%")result_dict = {"leftTop": c1, "rightBottom": c2, "Confidence": conf}return result_dictdef clip_coords(boxes, shape):# Clip bounding xyxy bounding boxes to image shape (height, width)if isinstance(boxes, torch.Tensor): # faster individuallyboxes[:, 0].clamp_(0, shape[1]) # x1boxes[:, 1].clamp_(0, shape[0]) # y1boxes[:, 2].clamp_(0, shape[1]) # x2boxes[:, 3].clamp_(0, shape[0]) # y2else: # np.array (faster grouped)boxes[:, [0, 2]] = boxes[:, [0, 2]].clip(0, shape[1]) # x1, x2boxes[:, [1, 3]] = boxes[:, [1, 3]].clip(0, shape[0]) # y1, y2def scale_coords(img1_shape, coords, img0_shape, ratio_pad=None):# Rescale coords (xyxy) from img1_shape to img0_shapeif ratio_pad is None: # calculate from img0_shapegain = min(img1_shape[0] / img0_shape[0], img1_shape[1] / img0_shape[1]) # gain = old / newpad = (img1_shape[1] - img0_shape[1] * gain) / 2, (img1_shape[0] - img0_shape[0] * gain) / 2 # wh paddingelse:gain = ratio_pad[0][0]pad = ratio_pad[1]coords[:, [0, 2]] -= pad[0] # x paddingcoords[:, [1, 3]] -= pad[1] # y paddingcoords[:, :4] /= gainclip_coords(coords, img0_shape)return coordsdef onnx_model_main(path):# onnxsession = onnxruntime.InferenceSession("last.onnx", providers=["CPUExecutionProvider"])start = time.time()image = open(path, "rb").read()img = np.array(Image.open(BytesIO(image)))# img = cv2.imread(path)# 图像处理img = img[:, :, :3]im = padded_resize(img)# 模型调度pred = session.run([session.get_outputs()[0].name], {session.get_inputs()[0].name: im})[0]pred = torch.tensor(pred)pred = non_max_suppression(pred, conf_thres=0.25, iou_thres=0.60, max_det=1000) # 大于百分之六十的置信度coordinate_list = []for i, det in enumerate(pred):det[:, :4] = scale_coords(im.shape[2:], det[:, :4], img.shape).round()for *xyxy, conf, cls in reversed(det):# 返回坐标和置信度coordinates = return_coordinates(xyxy, conf)coordinate_list.append(coordinates)# 坐标列表coordinate = sorted(coordinate_list, key=lambda a: a["Confidence"])# 用时duration = str((time.time() - start))if len(coordinate) == 0:data = {'message': 'error', 'time': duration}else:coordinate = coordinate[-1]x = coordinate.get('leftTop')[0]y = coordinate.get('leftTop')[1]w = coordinate.get('rightBottom')[0] - coordinate.get('leftTop')[0]h = coordinate.get('rightBottom')[1] - coordinate.get('leftTop')[1]point = f"{x}|{y}|{w}|{h}"data = {'message': 'success', 'time': duration, 'point': point}data.update(coordinate)print(data)return datadef drow_rectangle(coordinate, path):img = cv2.imread(path)# 画框result = cv2.rectangle(img, coordinate.get("leftTop"), coordinate.get("rightBottom"), (0, 0, 255), 2)cv2.imwrite("drow_rectangle.jpg", result) # 返回圈中矩形的图片print("返回坐标矩形成功")
调用方法:
coordinate_onnx = onnx_model_main("1.png")#需要识别的图片路径#返回内容#{'message': 'success', 'time': '0.5251204967498779', 'point': '260|90|59|56', 'leftTop': (260, 90), 'rightBottom': (319, 146), 'Confidence': 0.31054314970970154}##这个是根据返回信息在同级目录下生成一个加了框的图片drow_rectangle(coordinate_onnx, "1.png")#需要识别的图片路径
效果类似于这样子
附带个顶象的测试代码,基于selenium的:(这俩放一起就行,别问我为什么不是协议,问就是不会行了吧)
class Test():def __init__(self):option = Options()option.add_experimental_option('excludeSwitches', ['enable-automation'])option.add_argument('--disable-blink-features=AutomationControlled')self.driver = webdriver.Chrome(options=option)def __ease_out_expo(self, sep):if sep == 1:return 1else:return 1 - pow(2, -10 * sep)def generate_tracks(self, distance):"""根据滑动距离生成滑动轨迹:param distance: 需要滑动的距离:return: 滑动轨迹<type 'list'>: [[x,y,t], ...]x: 已滑动的横向距离y: 已滑动的纵向距离, 除起点外, 均为0t: 滑动过程消耗的时间, 单位: 毫秒"""if not isinstance(distance, int) or distance < 0:raise ValueError(f"distance类型必须是大于等于0的整数: distance: {distance}, type: {type(distance)}")# 初始化轨迹列表slide_track = [[random.randint(-50, -10), random.randint(-50, -10), 0],[0, 0, 0],]# 共记录count次滑块位置信息count = 30 + int(distance / 2)# 初始化滑动时间t = random.randint(50, 100)# 记录上一次滑动的距离_x = 0_y = 0for i in range(count):# 已滑动的横向距离x = round(self.__ease_out_expo(i / count) * distance)# 滑动过程消耗的时间t += random.randint(10, 20)if x == _x:continueslide_track.append([x, _y, t])_x = xslide_track.append(slide_track[-1])return slide_track# 顶象测试def dx_test(self):url = 'https://www.dingxiang-/business/captcha?utm_source=baidu1sem&utm_medium=%E4%BA%A7%E5%93%81%E8%AF%8D&utm_campaign=%E4%BA%A7%E5%93%81%E8%AF%8D-%E9%AA%8C%E8%AF%81%E7%A0%81&utm_term=%E9%AA%8C%E8%AF%81%E7%A0%81&e_matchtype=1&e_keywordid=317912325143&bd_vid=9281303599717237405'# Chrome浏览器self.driver.get(url=url)# //li[@class="item-2"]/h3self.driver.maximize_window()# 滑动到最底部js_button = 'q=document.body.scrollTop=500'# 执行js,滑动到最底部self.driver.execute_script(js_button)self.driver.find_element(By.XPATH,'//li[@class="item-2"]/h3').click()time.sleep(2)self.driver.find_element(By.XPATH,'//div[@id="dx_captcha_oneclick_bar-inform_3"]').click()time.sleep(2)self.download_yzm()distence = onnx_model_main('1.png')drow_rectangle(distence,'1.png')time.sleep(1)distence = int(distence['leftTop'][0])-30source = self.driver.find_element(By.XPATH,'//div[@id="dx_captcha_basic_slider-img-animated-wrap_4"]/span[1]')action = ActionChains(self.driver,duration=20)action.click_and_hold(source)a = 0time.sleep(1)for x in self.generate_tracks(distence):print(x)action.move_by_offset(xoffset=x[0] - a, yoffset=x[1])a = x[0]time.sleep(0.5)action.release().perform()input()# 网易测试def wy_test(self):url = '/trial/jigsaw' # Chrome浏览器self.driver.get(url=url)self.driver.maximize_window()js_button = 'q=document.body.scrollTop=500'# 执行js,滑动到最底部self.driver.execute_script(js_button)self.driver.find_element(By.XPATH,'//li[@class="tcapt-tabs__tab"][1]').click()time.sleep(2)self.driver.find_element(By.XPATH, '//button[@class="yidun_refresh"]').click()html = self.driver.page_sourcehtml = etree.HTML(html)url = html.xpath('//img[@class="yidun_bg-img"]/@src')[0]response = requests.get(url)print(url)with open('1.png', 'wb')as f:f.write(response.content)time.sleep(2)input(11111111111111)self.download_yzm()distence = onnx_model_main('1.png')drow_rectangle(distence,'1.png')time.sleep(1)distence = int(distence['leftTop'][0])-30source = self.driver.find_element(By.XPATH,'//div[@id="dx_captcha_basic_slider-img-animated-wrap_4"]/span[1]')action = ActionChains(self.driver,duration=20)action.click_and_hold(source)a = 0time.sleep(1)for x in self.generate_tracks(distence):print(x)action.move_by_offset(xoffset=x[0] - a, yoffset=x[1])a = x[0]time.sleep(0.5)action.release().perform()input()# 测试代码用到的下载图片的东西def download_yzm(self):js = "return document.querySelector('.dx_captcha_basic_bg > canvas').toDataURL('image/png')"image_data = self.driver.execute_script(js) # 执行js代码得到图片数据data = image_data.split(',')[1]image_data = base64.b64decode(data)with open('1.png', 'wb') as f:f.write(image_data)if __name__ == '__main__':a = Test()a.dx_test()# coordinate_onnx = onnx_model_main("1.png")# drow_rectangle(coordinate_onnx, "1.png")
可能出现的错误:
1、xpath定位错误
2 、action = ActionChains(self.driver,duration=20) 这里报错
3、self.driver.find_element(By.XPATH,‘xxx"]’).click()这里报错
4、 self.driver = webdriver.Chrome(options=option)这里报错
5、环境问题
解决方法:
1、自己改改嘛,挺简单的。
2、删掉duration=20就好了 或 升级下selenium版本。
3、改成find_element_by_xpath(‘xxx’) 或 升级下selenium版本。
4、指定下chromedriver.exe位置 或 放在python目录下。
5、仅仅是用的话笔者提供的示例的话,应该没啥问题吧…应该,实在不行加我qq问吧(请我吃根棒棒糖就可以了!我这里棒棒糖仅需5毛!!!)
模型下载:
1、csdn下载:通用滑块验证码识别模型(我设置的0积分,如果收费了记得给我说下,或者直接加我,问我要也可以)
2、百度云下载: /s/1Nt2Z2Pu45RaQBs4dsS2kKg?pwd=82xg
提取码:82xg (老慢了)
3、github:暂无
百度云下载方式更新在了github
github: /Bump-mann/simple_ocr
测试:http://121.4.108.95:8000/index/
参考:
1、深度学习训练滑动验证码(Yolov5)
2、【Yolov5】1.认真总结6000字Yolov5保姆级教程(全新版本.06.28)
---------------------------------------笔者碎碎念---------------------------------------
本来吧,是想先出旋转验证码的,代码都写好了,但是大家都想看滑块的,就先出滑块的啦…
开始做的时候想着就图片处理嘛,蛮简单的,结果…一家一个类型,参数没法写死,不同的图片不同的算法效果不稳定,全是坑,被逼无奈改成写模型了
什么?你问我代码能通杀吗?能,当然能,前提是你数据量够
观众老爷:你这不是欺负老人吗?标题党系内!
笔者:阿巴阿巴,如果催的人多的话,我会持续更新的,做到真正的通杀
本代码的数据标注都是笔者个人一个框一个框标注的,呜呜呜,我也想对接打码平台啊,穷啊